Quality Monitoring And Analysis Of Steel Products Based On Real-time Data Stream

With the development of computer hardware and sensor networks, data stream is widelyused in various fields as a new data form.As the characteristics of data stream are differentfrom the traditional static data,traditional data mining methods are no longer suitable for datastream mining, which becomes a new research topic in the field of data mining,and hasbrought great challenges to research and industrial control fields.By analyzing the characteristics of real-time data streams and related technologies,thisthesis introduces some typical clustering algorithms that currently exist in theliterature,analyzes their advantages and disadvantages as well as the applicableenvironment.Finally,this thesis shows the application of the real-time data stream through onespecific examples.It proposes a method to do product quality supervision based on real-timedata analysis, aiming at the common quality analysis and supervision problem in the processof steel production. Combining with real-time database and relational database, the methodperforms a real-time detection of steel production in both real-time supervision and off-linetracing. The main work is the implementation of on the collection of real-time data forreal-time databases, and the analysis schema given both real-time and relational data. Themethod proposed has been applied in several production lines of a steel factory and performedwell quality management.