Abstract: To solve system integration and data management problems in the textile manufacturing process, through the textile manufacturing technology process, the massive data of each process is analyzed, and the inefficient convergence phenomenon of textile information between production planning layer and workshop manufacturing layer is studied. On the basis of the original system data, and text data of raw materials, sensor, yarn defect detection image data, and so on, a three layer textile big data storage system based on Hadoop is built. And then, through using theoretical methods of D-S evidence and incremental clustering, the technical difficulties that are multi-source textile data fusion is designed, the appropriate algorithms and models are proposed, and functions of the system are designed and implemented. Through the test, the results show that the system we designed have realizes the effective information link between planning layer and production layer by the correlation between data, solves the ‘information island’ problem, and can provide a new method for real-time online detection of fabric quality in big data environment.

Key words: big data, data fusion, manufacturing execution system, textile industry