Laser deep penetration welding intelligent monitoring system and method based on plasma electric signals

Abstract

The invention discloses a laser deep penetration welding intelligent monitoring system and method based on plasma electric signals. The system comprises three portions, namely, a laser welding subsystem, a motion control subsystem and an electric signal acquisition subsystem, so that the quality judgment of weld joints is realized. The method comprises the following steps that the electric signalsof the laser plasma are measured on line, and the electric signal voltage data are acquired in real time; the denoised signals undergo ensemble empirical mode decomposition; sample data of each weldjoint is selected; the node number of an input layer, a hidden layer, a sum layer and an output layer of the probability neural network to be established is determined; by taking the training sample data as input of the probability neural network, a probability neural network model is established, probability neural network training is performed, the judgment result of the probability neural network is compared with the actual weld joint state, and the probability neural network model is verified; and the prediction accuracy of the probability neural network model is verified. According to thelaser deep penetration welding intelligent monitoring system and method, the welding state can be identified in a high-precision mode, the feedback control of the welding quality can be realized, andthe operation cost is greatly reduced.

Description

technical field [0001] The invention relates to the technical field of welding quality detection, in particular to an online intelligent monitoring system and method for laser deep penetration welding based on plasma electrical signals. Background technique [0002] With the active penetration of laser processing technology into the traditional manufacturing industry, the welding method using laser as the energy carrier has been widely used in the industrial field. Welding occupies an irreplaceable important position in the manufacture of key components of precision instruments. In order to avoid complex non-destructive testing after welding and reduce industrial production costs, online quantitative detection of welding defects based on sensor information has become one of the most attractive research topics. The current research generally uses technical means such as visual sensing, acoustic emission, and spectral diagnosis. However, due to the huge image information gene...

AI Technical Summary

Problems solved by technology

The current research generally uses technical means such as visual sensing, acoustic emission, and spectral diagnosis. However, due to the huge image information generated by the visual sensor, the processing speed is not fast enough, and it is difficult to use it in real time; and the two sensing technologies of acoustic emission and spectral diagnosis, There are still deficiencies in the judgment of defect types

Images