Abstract: With the development of automatized and flow-line welding production,high requirement on quality control and management has been put forward.In order to discover latent rules and tendency and provide decision support from large historical production data,data mining is introduced to welding field and the management platform is developed based on on-line monitoring platform and database.At the same time,the data is pre-processed and data market is built so as to obtain valid data mining model.Control chart is adopted to control welding production and wasters and decision tree is used to analyze factors leading to defects,which obtain significance for welding production.