Study On The Application Of Kalman Filtering State For Seam-tracking

As one of the most important methods of machining craftwork in industry, welding jointing shows its weightiness increasingly with the development of our country's modernization construction. Not only the study on and the development of welding tracking are the precondition of carrying out the automatic and intelligent welding, but the important means of insuring welding quality, raising productivity and improving work conditions.Vision sensing method is one of the most important sensing methods available in real-time welding seam tracking, which often extracts the welding seam information after obtaining the welding images through reducing the effect of noise and uses it to accomplish the welding tracking. Aiming at the inherent defects such as the intrusion of intensive arc and the error brought by the preposed sensor, the welding tracking method based centroid detecting is proposed.Differing from the traditional method, the welding tracking method based centroid detecting use the centroid of welding pool image as the detecting information through analyzing the whole welding pool image and extracting the character of the image instead of extracting the welding seam information directly. Through the establishment of the state equation and the measurement equation, the Kalman Filtering can be applied to the system to reduce the effects of the process and measurement noise on the centroid. Meanwhile, the estimation of the current state of the centroid can be obtained, which often is used to deduce the actual welding seam information. Then, the necessary control signal to the control system can be obtained. And the prediction function of Kalman Filtering is proposed to provide the prediction of the centroid of the next time, which can be used as the reference of the tracking system. In this way, the error of the tracking system brought by the preposed sensor is reduced and a better performance of tracking control can be carried out.The state vectors of the centroid is extracted through analyzing the movement of torch in the process of welding and the state equation and the measurement equationbased the centroid of welding pool image is established. A simulation experiment is carried out to demonstrate the feasibility of the application of the Kalman Filtering based the estimation of state to the welding process. Based on the welding experiments, the effect of the application of Kalman Filtering is verified and the problems derived from the application process are analyzed. After a great deal of experiments, the resolution to the problems mentioned above is proposed, which analyzes the data of experiment centroid , and the better choice of initial value is achieved.The process noise of the system and the measurement noise are eliminated effectively by Kalman Filtering, which is demonstrated by the results of welding experiments. Based on sufficient experiments, the effects of Kalman Filtering have be verified and analyzed in the form of errors and covariance. The value of the method taking the centroid as the detected information to the tracking system has been demonstrated.The factors resulted to the error brought by the preposed sensor have been further analyzed theoretically Based on the one-step predict function of Kalman filtering, a new method for solving the the error brought by the preposed sensor was put forward finally.