Automatic Classification and Coding of Low-Level Events in Nuclear Power Plants Based on Machine Learning

• Published in: 2021 IEEE 24th International Conference on Computer Supported Cooperative Work in Design (CSCWD) pp. 1003 - 1008
• Main Authors: Zhang, Lian, Cai, Hankun, Yang, Meng, Ying, Qiang
• Format: Conference Proceeding
• Language: English
• Published: IEEE 05.05.2021
• Subjects: Automatic Classification Coding; Bert; Classification Algorithm; Classification algorithms; Computational modeling; Deep learning; Encoding; Machine Learning; Machine translation; Nuclear Power Plant; Status Report; Task analysis
• DOI: 10.1109/CSCWD49262.2021.9437861

At present, Experience Feedback Systems have been established in all currently operating and under construction nuclear power plants, and as such, a large number of low-level events have been collected. Due to human and material constraints, these low-level events have not been well utilized. The main problem is that the number of low-level events is so large that it is impossible to code all events manually. In this paper, through the use of seven classification learning algorithms, automatic coding research on low-level events of nuclear power plants is carried out to demonstrate the feasibility of applying a classification learning algorithm to the field of nuclear power text processing. This paper first introduces low-level events in nuclear power plants, low-level event trend coding, and the classification algorithms based on machine learning and deep learning used in this study. Then, comparative experiments are designed and the results of the experiments are analyzed in detail. Finally, based on the results of this study, suggestions for the use of artificial intelligence in the field of nuclear power text processing are proposed. This study provides a reference for the application of machine learning in the field of nuclear power texts processing.