Study On The Quality Inspection & Prediction Of Rail Flash Butt Welding

Rail welding is the key step of high-speed railway laying project.Alloyed rail steel applicated in the rail engineering in China is of high strength and good wearability. Because it has not a good welding capacity, and the inspection method of rail weld quality is not perfect, so there is some hidden danger in the rail line, therefore it's very important to do research about on-line quality prediction of flash butt welding.A series of research findings about on-line quality prediction of flash butt welding were obtained by welding institude of Southwst Jiaotong University with artificial neural network (ANN),but there are still some shortages in the system. Firstly, there are some connection and compatibility problems of two systems, because the software platform of data acquisition system, coded with the soft of Labview of NI?, is differrent form data analysis system, coded with the soft of Visual Basic of Microsoft?. Secondly, using the impacted quality as judicial standard of welding quality is not perfect, because the impacted quality is not only relation to welding quality, but also to others, therefore it's more rational using the area of grey-spots. Aiming at these problems, new software is coded with the soft of Visual Basic, which is of the functions of high-speed data acquisition and welding quality prediction.The application indicated that welding process parameters can be recorded exactly and detaily by new system with a good man-machine interface,and the area of grey-spots can be predicted with ANN prediction model.The requirements of the real-time ability and the accuracy of prediction can also be meet by the new system.