A Two-Stage Machine Learning-Aided Approach for Quench Identification at the European XFEL

Lynda Boukela; Annika Eichler; Julien Branlard; Nur Zulaiha Jomhari

Home/Archive/arXiv

arXiv · 2407.08408

A Two-Stage Machine Learning-Aided Approach for Quench Identification at the European XFEL

Lynda Boukela·Annika Eichler·Julien Branlard·Nur Zulaiha Jomhari

View original source ↗Read source document ↗Export metadata ↓

Abstract

This paper introduces a machine learning-aided fault detection and isolation method applied to the case study of quench identification at the European X-Ray Free-Electron Laser. The plant utilizes 800 superconducting radio-frequency cavities in order to accelerate electron bunches to high energies of up to 17.5 GeV. Various faulty events can disrupt the nominal functioning of the accelerator, including quenches that can lead to a loss of the superconductivity of the cavities and the interruption of their operation. In this context, our solution consists in analyzing signals reflecting the dynamics of the cavities in a two-stage approach. (I) Fault detection that uses analytical redundancy to process the data and generate a residual. The evaluation of the residual through the generalized likelihood ratio allows detecting the faulty behaviors. (II) Fault isolation which involves the distinction of the quenches from the other faults. To this end, we proceed with a data-driven model of the k-medoids algorithm that explores different similarity measures, namely, the Euclidean and the dynamic time warping. Finally, we evaluate the new method and compare it to the currently deployed quench detection system, the results show the improved performance achieved by our method.

Explore related subjects

physics.ins-det cs.AI

Keep this discovery

Explore connections, maps & timelines

BibTeX RIS

Lynda Boukela, Annika Eichler, Julien Branlard, Nur Zulaiha Jomhari. 2024-07-11. A Two-Stage Machine Learning-Aided Approach for Quench Identification at the European XFEL. https://arxiv.org/abs/2407.08408

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related papers

arXiv2026

TCAD + Allpi$\text{x}^2$ Simulation study of MALTA2, a Depleted Monolithic Active Pixel Sensor for future tracking

In this work, a hybrid simulation framework combining TCAD and Allpi$\text{x}^2$ is presented to investigate the sensor properties of MALTA2, a depleted monolithic active pixel sensor designed for future tracking. The study starts from 3D modeling and transient simulations in TCAD, with generic doping profiles and simple well structures. The resulting doping profiles and electric field are extracted and fed into Allpi$\text{x}^2$ for high-statistics Monte Carlo simulations in both DUT-only and full-telescope mode. Simulations reveal a strong dependence of sensor performance, specifically the detection efficiency and cluster size, on the doping concentration of the N-type blanket at the sensor surface. The doping concentration is then optimized by comparing simulations with measurement data. The active depth of the depleted region of the MALTA2 sensor is estimated in both simulations and measurements using a grazing angle method, in which the sensor is positioned at various inclinations relative to the beam, covering angles from 0 to 60 degrees. Excellent agreement on active depth is obtained with the optimal doping concentration, showing a deviation of 2\% from the measured value at a threshold of 450\,$\text{e}^-$. Consequently, the framework offers a generic toolkit for sensor studies without requiring proprietary information.

physics.ins-det↗

arXiv2026

A compact fibre-coupled active beamstop for fast time-resolved monitoring of synchrotron X-ray beams

We have developed a compact fibre-coupled active beamstop for simultaneous interception and time-resolved monitoring of the direct X-ray beam in synchrotron diffraction experiments. The complete beamstop tip has a maximum transverse dimension of 1 mm and incorporates a 700 $μ$m-wide Ce:YAG scintillator. The scintillator converts the intercepted X-rays into visible light, which is transported through a multimode optical fibre to a silicon photodiode at the bottom of the blade that is connected to a dedicated fast amplifier electronics. Tests with a pulsed laser yielded a system decay time of 69.4 ns, consistent with reported luminescence decay times for Ce:YAG. Measurements at the P14.EH2 end station of PETRA III demonstrated the detection of individual bunch signals separated by 192 ns in the 40-bunch operating mode. The active beamstop also provided an in-situ timing signal for synchronising an experimental sequence with the approximately 190 $μ$s transmission window of a mechanical X-ray chopper. The device therefore combines a compact obstruction of the direct beam with fast monitoring of X-ray timing and flux during diffraction data acquisition.

physics.ins-det↗

arXiv2026

Charge-Carrier transport simulations in diamond detectors with electric-field-dependent mobility and charge-collection-distance-based trapping

Diamond detectors are attractive for operation in harsh radiation environments because they combine radiation tolerance, fast signal formation, and low leakage current. Realistic detector-response simulations require an accurate description of charge-carrier mobility and trapping, which determine both signal amplitude and timing. In this work, we extend \allpix{}, a modular end-to-end detector simulation framework, with diamond-specific transport models. The implementation includes field-dependent mobility parameterizations for electrons and holes and an effective trapping model based on the charge collection distance (CCD), providing a detector-level interface to material quality and radiation-damage measurements. The mobility description is validated in the negligible-trapping limit using single-crystalline CVD diamond by comparing simulated drift velocities and transient-current signals with published reference data. For polycrystalline CVD diamond, the CCD-based trapping model is evaluated using experimentally measured CCD values and compared with laboratory transient-current-technique waveforms. The simulations reproduce the measured drift-velocity behavior in scCVD and the reduced charge collection and degraded transient response observed in pcCVD. The presented implementation enables detector-level studies of charge collection, pulse formation, and timing performance in diamond sensors using experimentally accessible transport and trapping parameters, and provides a practical framework for simulation-driven detector development and radiation-damage studies.

physics.ins-det↗