Machine Performance and Condition Monitoring Through Data Mining and Database Optimization

Engineering dataset has growing rapidly in size and diversities with data acquisition technology development in recent years. However the full use of the datasets for maximizing machine operation and design has not been investigated systematically because of the complexity of datasets and large scale of data amount. This also means that traditional statistical data analysis methods are no longer the efficient methods to obtain useful knowledge from the dataset. Therefore this study will focus on applying more advanced data minting technologies and optimizing database systems so that more accurate and efficient knowledge can be extracted from engineering datasets for machine performance and condition monitoring. For the first year study, a full understanding engineering dataset has been obtained based on the condition monitoring activities in the DERG laboratory.
In the mean time a review has also conducted on different techniques such as Neural networks, clustering, genetic algorithms, decision trees and support vector machines that are used widely for data mining. Moreover, to evaluate the effectiveness of using the techniques, a prototype
database has been developed and applied with the methods. The results obtained from this evaluation study have shown that the data mining can be efficient to obtain information for condition monitoring. But more work on developing methods to optimize both the parameters of using
the methods and the database organization.