ABSTRACT
Material testing system in our country mainly includes test equipment, test process
monitor, data acquisition. Material test management is relatively simple. Experts discuss
and analysis the test results. Intelligent materials science, which combined with the
computer technology, is relatively backward. In the field of materials science, decision
support theory in some foreign companies have been studied and implemented, in some
respects, the effort is remarkable. Based on the analysis of the materials testing industry
status and the actual project requirements, using the object-oriented methods, intelligent
data analysis, data mining techniques in this paper, we design the material testing
system, research and apply decision support theory in this system.
In the system design, considering the distributed web-based material test, for the
different material testing requirements and measurement parameters, the database
design can configure the test flexibly, store the test data, the overall design achieves
material test data acquisition and test control. We proposed a detailed design plan for the
decision support system, design the data warehouse with star schema and snowflake
structures combined, in order to provide an effective data source for decision support.
Make the meta-data configuration management, realize the data extraction,
transformation and load from the database to the data warehouse clearly and definitely.
In the theoretical research, we focus on the data processing and data modeling
research; optimize time series data mining algorithm, simplify the test sample data sets
effectively and easy to store. Provide a high-quality data for data modeling; make the
material analysis based on the valid data sets. Research and sum up the neural network
applications in the materials science. Using object-oriented methods to design the neural
network, the system provides modeling method to the outside. Eventually, the system
will be able to do data analysis, data modeling, and data mining, present the results to
the users. The overall design and research meet the needs of materials science, have
realistic meaning.
Key Words: Material Testing System, DSS, Data Mining, system design,
Time Series, neural network