Nuclear reactor internal component surface roughness assessment method based on convolutional neural network

Abstract

The invention discloses a nuclear reactor internal component surface roughness assessment method based on a convolutional neural network. The method comprises the following steps of S1, collecting video data; S2, obtaining image data; S3, dividing the image data into a training data set and a test data set; S4, inputting the training data set into the convolutional neural network for training, obtaining a feature identification model and outputting initial identification features; S5, classifying the initial identification features according to roughness, thereby obtaining roughness levels; S6, inputting the test data set into the feature identification model for test, and outputting secondary identification features; S7, judging whether the secondary identification features are consistentwith the roughness levels or not; and S8, displaying the roughness levels on a human-computer interaction interface. According to the method, the problem that in the prior art, the manpower investment is high and the efficiency is low due to artificial detection and assessment, and the material surface roughness detection and assessment timeliness and continuity requirements cannot be ensured issolved.

Description

technical field [0001] The invention relates to the technical field of the nuclear industry, in particular to a method for evaluating the surface roughness of nuclear reactor internal components based on a convolutional neural network. Background technique [0002] Before 2020, my country will build another 32 million-kilowatt-class nuclear power units, including those already built in Qinshan, Daya Bay, and Ling'ao, with a total installed capacity of more than 40 million kilowatts. Regular inspection and maintenance of in-service equipment in high-radiation underwater environment of nuclear power plants is an important guarantee for the safe operation of in-service nuclear power plants, and it is also a dangerous, hard and time-consuming job, which needs to solve the problems of high radiation dose and underwater operation feasibility. [0003] Judging from the current development situation in our country, to improve the safety and reliability of nuclear implementation, red...

AI Technical Summary

Problems solved by technology

[0010] Aiming at the above-mentioned deficiencies in the prior art, the present invention provides a method for evaluating the surface roughness of nuclear reactor internal components based on a convolutional neural network that saves manpower input, high efficiency, high real-time performance and continuity, and solves the problems of the prior art. Existing manual detection and evaluation lead to large manpower input and low efficiency, and cannot guarantee the real-time and continuous requirements of material surface roughness detection and evaluation

Images