Papers for

industrial sensor network teams

Papers whose findings have a practical use for this group, as judged from the abstract. Open a paper to read what it means in practice.

Foundation model improves sensor data use for particle accelerators

SensOFormer: Foundation Models for Particle Accelerators

Abstract: Large-scale scientific facilities, such as particle accelerators, are complex systems composed of thousands of interacting components that must operate collectively to support diagnostics and control, and ensure safe, reliable operation. These facilities rely on extensive sensor networks that generate large volumes of heterogeneous, incomplete, and noisy telemetry, often sampled at different rates across subsystems. In this paper, we adopt the foundation-model paradigm to learn general-purpose sensor representations from historical accelerator data through self-supervised pretraining. We introduce SensOFormer, a denoising masked transformer with a Perceiver-style encoder-decoder architecture designed to handle variable sensor sets, accommodate heterogeneous sampling rates, and learn robust representations from noisy measurements. The model is pretrained on multiple particle-accelerator datasets and transferred to downstream tasks, including missing-data imputation, anomaly detection, cavity identification, and fault identification. Through extensive evaluations and comparisons, we demonstrate that a single self-supervised model can learn reusable representations for diverse operational tasks in large-scale scientific facilities. Project page: https://mahindrautela.github.io/sensoformer.

Sun 1 NovMachine Learning
The gist
Particle accelerators use many sensors that produce lots of noisy and incomplete data, making it hard to monitor and control the machines. The authors created SensOFormer, an AI model that learns from past sensor data to understand and clean this information. SensOFormer can work with different sets and speeds of sensors, and it helps with tasks like filling in missing data and detecting faults. They trained this model on several accelerator datasets and showed it can handle many problems using one general approach.
Open → 2611.02031v1