| To achieve automatic and intelligent welding process,the first problem to be faced with is to realize automatic tracking of weld seam.To solve this problem,the key lies in the method of detecting weld seam center position parameters.Current pulsed eddy current detection,ultrasonic detection and X-ray detection method without loss is frequently seen as common methods.Compared to ultrasonic detection and ray detection,pulsed eddy current detection is of low cost,simple structure,showing a stronger anti disturbance against the dust in the process of welding,arc light and the background light.What's more,it does not require a specific coupling agent and conductive media.This paper presents a research on automatic tracking of weld metal pulsed eddy current nondestructive testing based on the first design of pulse tracking system of eddy current weld.Then introduces the extraction method of weld center parameters,then processing algorithm research of weld center.This paper could be mainly divided into the following aspects:(1)The weld tracking technology at home and abroad are briefly introduced.Then the development and application of eddy current sensor are described.According to the development trend of the eddy current sensor,the pulsed eddy current detection of weld,and through the test and analysis for the feasibility of automatic seam tracking.(2)The design of pulse eddy current weld tracking system,which is mainly divided into the implementation of welding device and the welding control device design.The former focuses on argon arc welding of various components in the system and the three-dimensional design movement of the console;While the latter on the pulsed eddy current testing of weld center,five axis driving circuit design,software interface design and the design of seam tracking system.(3)In the presence of noise in the collected signals,this paper uses the pulsed eddy current signal acquisition method of weld center and then make the signal synchronization,then use the median filter and the value of arithmetic average method for signal denoising,the denoising signal after differential processing and peak voltage the peak time of the weld center.(4)There are many kinds of interferences for the seam tracking process,seriously affectthe accurate extraction of weld position information of the status quo,to carbon steel plate butt welding test for the object,firstly by means of pulsed eddy current nondestructive detection of weld position peak and zero crossing time,on this basis,further the system state optimal estimation method of Kalman filter the welding center from the state of welding correction.The test results show that the Kalman algorithm can effectively eliminate the interference noises,optimization of pulsed eddy current extraction of information,improve the seam tracking stability.(5)In order to avoid the influence from the uncertainty of the noise characteristics on seam tracking,Kalman filter is used to optimize the algorithm of RBF through RBF neural network,based on Kalman filter parameters as network input,the filtering error as the output of the network so as to establish RBF neural network training process,and then using the outputs of the trained RBF neural network to modify Kalman filter value,compensating the filtering error in the center of the weld.The test results show that Kalman filtering algorithm using the optimized tracking accuracy is increased by 34.3% compared to that before optimization,and can reduce the influence of noise on the measurement data,correct the pulsed eddy current measurement data,obtaining more precise position of the weld center and the seam tracking accuracy is also enhanced. |
PDF Full Text Request