Research On Identification Method Of Pipeline Weld And Design Of Weld Ultrasonic Scanning Equipment

Industry pipeline is the basic equipment of modern industry, but because of its installation and application conditions, there are different degrees of defects in pipeline welding line. So it is very important to test the pipeline.In the process of pipeline inspection, the reliability of the test results is highly dependent on the responsibility of field staff and technical levels, due to strict-installation requirements and stringent-application conditions, different degree of defects inevitably exist in butt girth welds of pipeline. Therefore, developing an automatic detection device of the pipeline weld with a simple structure, good stability, convenient installation,which can at the same time reduce the dependency on field staff has a significant practical meaning.The tracking of welding seam is the precondition of ultrasonic inspection to pipeline, and the key is the recognition of welding seam and the comfimation of position. In this paper, based on the difficult problem of welding seam recognition, combined with the application of the welding seam scanning device, this paper firstly put forward the method of welding seam recognition method with artificial navigation marks. By adding an artificial navigation mark on the weld to distinguish the weld and the base metal region, the image processing can be obtained by using the image processing to extract the offset weld binarization image. Then based on the idea of modularization, combined with the identification method for design of the scanning device of the motion control system, design and selection of main controller, power supply circuit, drive circuit, camera module, remote control module, the hardware circuit of the control system is built, and the software design of control system. After many tests, the scanner meets the requirements of industrial inspection.Secondly, A group of contrast images shows that the gray difference and the layer of weld area and the surrounding area of the base metal are not obvious, and the gray scale gradient of the welding seam edge is not evident. Therefore, it is incapable to identify the welding seam position accurately by simple image-edge detection and gray threshold segmentation. Then, according to the texture characteristic parameters of the welding seam image, the recognition method of "one side learning, one side correction" is proposed, The method has high successful rate and can effectively determine the position of the welding seam.Then,on the basis of feature parameters of welding image texture, combining with BP artificial neural network to set up the neural network model, the four-dimensional texture characteristic parameters as input, design the network input layer, output layer and transfer function, training function, the hidden layer. It can be known from the experiment that the network model has good convergence effect, which further improves the successful rate of weld recognition.Finally, the improvement scheme of the hardware of the scanning device is proposed. Based on the method of BP neural network and image texture, the improved algorithm is designed, which can be used as a classifier to recognize the weld seam.Which lays the foundation for the further improvement of the scanning device.