Study on an abnormal temperature detection method for CAFe facility

Abstract

Beam availability is an important indicator of the operational stability and safety of the accelerator. As the front-end prototype of CiADS (China initiative Accelerator Driven System), the CAFe (Chinese ADS Front-end Demo Linac) accelerator is used to verify the feasibility of the key technologies of CiADS, which plays an important role in the development of nuclear waste safety treatment technology. The accelerator is often affected by uncontrollable factors during operation, triggering false temperature alarms from the BPM (Beam Position Monitor), causing the accelerator machine protection system to cut the beam, significantly reducing the availability of the beam. In addition, the prevailing method of manually setting the machine protection system threshold increases the beam conditioning time and the probability of false beam cuts. Therefore, it is urgent to study and discriminate the anomalies that cause beam cutting in order to reduce the beam cutting times caused by false alarms. In this paper, a BPM temperature anomaly prediction model based on Isolation Forest is developed, and the historical BPM temperature data from the CAFe accelerator is pre-processed using the Kalman filter to eliminate the influence of noise. The experimental results show that the scheme can accurately predict the abnormal state of the accelerator to remove the anomalous values, thereby avoiding false interlocks, and provide an effective interlock reference for the threshold setting of the machine protection system, thereby improving the beam availability.