Nuclear power plant low-level event automatic trend analysis method based on machine learning

Abstract

The invention provides a nuclear power plant low-level event automatic trend analysis method based on machine learning. The nuclear power plant low-level event automatic trend analysis method comprises the following steps of 1, establishing a nuclear power plant low-level event professional lexicon and performing automatic word segmentation; 2, realizing automatic classification coding of low-level events of the nuclear power plant by utilizing machine learning and deep learning algorithms; 3, performing automatic identification and early warning on the abnormal trend of the low-level event of the nuclear power plant based on the coding result; and 4, analyzing the root cause and formulating continuous tracking and evaluation for correcting the abnormal trend of the action. According to the machine learning-based automatic trend analysis method for the low-level events of the nuclear power plant provided by the invention, automatic coding of the low-level events of the nuclear power plant can be realized, and an abnormal trend and a common problem are automatically identified and positioned through coding-based trend analysis; and continuous tracking and evaluation of the abnormal trend subjected to root cause analysis are realized.

Description

technical field [0001] The invention relates to the technical field of nuclear power, in particular to an automatic trend analysis method for low-level events in nuclear power plants based on machine learning. Background technique [0002] Behind the high-level events (including operational events and internal events) in nuclear power plants are hidden a large number of low-level events and near-miss events, and the high-level events, low-level events and near-miss events are likely to have the same root cause or incomplete barrier. The trend analysis of low-level events and near-miss events helps to detect adverse trends in nuclear power plants early, identify incomplete barriers, take corrective actions before more serious events occur, eliminate trends, and achieve optimal management and focus on prevention It also provides managers with a new management tool. [0003] One of the basic tasks of trend analysis is coding, that is, categorizing events. Coding is labor-int...

AI Technical Summary

Problems solved by technology

[0005] It can be seen that when nuclear power plant experience feedback staff analyze the trend of low-level events in nuclear power plants, they need to manually code a large number of low-level events in nuclear power plants, and then search for abnormal trends by manually combing statistical data of hundreds of classification codes. Cause analysis and recording of corrective actions, there are problems such as huge labor costs, manual coding is easy to confuse, abnormal trend search is missing, cause analysis is insufficient, and corrective actions are not targeted.

Images