Kalman Filtering Of Magneto-optical Imaging For Micro-gap Seam Tracking

With the wide application of high-precision welding techniques, the importance of detection and tracking of squared groove and tightly-butted micro-gap weld seam is becoming increasing prominent. Accurate weld seam detection is the key to ensure weld quality and realize weld automation. However, traditional detection methods fail to extract effective information of micro-gap weld seam position and to recognize weld seam deviation. This paper thus proposes an innovative method for weld seam detection based on magneto-optical imaging.MOI sensing technology is a new technology to realize the visualization of magnetic field by detecting the amplitude of polarized light based on Faraday magneto optical effect. With regard to the micro-gap weld seam of tightly-butted low-carbon steel, a stimulating magnetic field is put on the work piece and magneto optical images of weld seam are obtained through magneto-optical sensors. Obvious transitional features of weld seam position can be observed in the magneto-optical images due to the distribution differences of the magnetic fields enacted on two different work pieces. Weld seam grey level distribution features in the magneto-optical images are investigated and with the integration of image processing methods, weld seam transitional zone is decided. The central point in the transitional zone is taken as the feature of weld seam position. A mathematical model about weld seam feature parameters and weld seam deviation is established to calculate the measurement information of weld seam deviation. Testing experiments with different welding conditions are carried out to verify how magneto-optical imaging sensing technologies can help to effectively obtain weld seam position information.During weld seam tracking process, there are various kinds of noise interferences which greatly affect the accurate extraction of weld seam position information. Therefore, an optimum estimation method for system status based on Kalman filter is proposed to make accurate prediction on weld seam deviation. Firstly, system status equation and weld seam position measurement equation are set up based on weld seam position parameters. The optimal prediction of seam offset could be obtained through the Kalman filtering algorithm under the least squares condition. Secondly, considering that the system processing noise and the measurement noise might be colored noises, an extended Kalman filtering algorithm is established through state amplification method and measurement amplification method to realize status optimum estimation under colored noise. Experimental results show that extended Kalman filter algorithm can effectively eliminate noise interference and improve the stability of weld seam tracking.It is difficult to obtained prior to actual welding wathe accurate information of the system processing noise and measurement noise statistical characteristics during welding seam tracking process The impact of the uncertainty of the noise statistical characteristics on Kalman filtering is researched through comparing the Kalman filtering results of different noise variance value. The neural network and the Kalman filter algorithm were combined, and a Kalman filtering algorithm compensated by neural network is investigated. A BP neural network is established, using Kalman filtering parameters as its input and filtering error as its output. The Kalman filtering result were modified by the BP network and the filtering error caused by the noise statistical characteristics was compensated. The experimental results verify the effectiveness of the method in improving the stability of filtering.