Abnormality detection method and device for beam position monitor based on auto-encoder

Abstract

The invention discloses an abnormality detection method and a device for a beam position monitor based on an auto-encoder. The method comprises: obtaining turn-by-turn beam position data of the beam position monitor; inputting the turn-by-turn beam position data as raw input data into a pre-trained auto-encoder model after preprocessing, processing and outputting reconstructed input data by the auto-encoder model; calculating a reconstruction error between the raw input data and the reconstructed input data; and determining whether the corresponding beam position monitor is abnormal by comparing the reconstruction error with an abnormality threshold, wherein the beam position monitor whose reconstruction error exceeds the threshold is determined to be in an abnormal or fault state. The abnormality detecting method and the device for the beam position monitor based on the auto-encoder solves the problems of abnormality of beam position monitors of storage rings and automatic identification of faults, improves the efficiency and reduces the research and development cost, and the method has high accuracy, high reliability and high practicability.

Description

technical field [0001] The invention relates to the technical field of storage ring beam current diagnosis and control, in particular to an autoencoder-based beam position monitor abnormality detection method and device. Background technique [0002] A storage ring is a circular accelerator widely used in synchrotron radiation sources and circular colliders. The beam position monitor is an important beam diagnostic component of the storage ring. In the storage ring, the beam trajectory of the entire storage ring is monitored by the beam position monitors distributed at different positions. The abnormal data caused by the failure of the beam position monitor will seriously affect the control of the beam trajectory and the stability of the beam in the storage ring, and will also affect the results of the model calibration of the storage ring. Therefore, identifying and detecting faulty BPMs and eliminating abnormal "bad" data generated by faulty BPMs has become an indispensab...

AI Technical Summary

Problems solved by technology

Since this method regards the fault identification of the beam position monitor as a binary classification problem based on the mechanism of shallow machine learning, there are the following problems: a large amount of time-consuming manual fault signal feature extraction is required, and the beam position monitoring The efficiency of the application on the storage ring with a large number of detectors is not high: and some potential hidden fault features are difficult to separate and extract through the singular value decomposition (SVD) method, resulting in low accuracy of beam position monitor fault identification

Images