Visualization and Monitoring Dynamic Water Levels of Steam Generators Based on Deep Learning

Abstract

In a nuclear power system, the steam generator is the most important part of the
second loop, and the normal monitoring of the internal water level is a strong
guarantee for the safe operation of the steam generator. In this study, a two-loop
steam generator experimental platform is mounted on a six-degrees-of-freedom
platform to fully imitate the working condition of the steam generator and visualize
the internal water level by opening a window on the steam generator. The level
monitoring is achieved by a deep learning recognition network (YOLO-v7), which is
improved by adding an attention mechanism to optimize the network to achieve better
level recognition and improve the recognition accuracy of the network by about 0.1.
By finding the relationship between the longitudinal pixels of the image and the actual
height, the recognition of the water level position in the image is mapped to the actual
water level height. By photographing the variation of water level in the steam
generator and using the recognition network to monitor the real-time water level
change, this method can achieve a fast dynamic response to the water level change,
which is a reliable and powerful guarantee for the level regulation of the steam
generator.

Keywords: steam generator, Level monitoring, deep learning, YOLO-v7, Marine conditions