With the application of transformer facility status analysis system, large amount of history datum accumulated. People push more and more demands on decision-making based on history datum. These brand-new demands, companied with the development of data warehouse technique, make it necessary to build a warehouse on transformer facility status analysis.Be aware of the necessity and feasibility of the data warehouse technique, we have designed and implemented a data warehouse of transformer facility status analysis system. The system is comprised of database, operational data store and data warehouse.Operational data store is a data storage technique which is between database and data warehouse, which is an architectural construct that is subject oriented, integrated, volatile, current valued, and contains detailed corporate data. ODS can recover the shortcoming that DW is hare to deal with real-time data.We often store the data in the warehouse as materialized view to speedup query processing on large amount of data. These views need to be maintained in respond to update in the source data. We show that the warehouse views can be made self-maintainable with the auxiliary views, which derived from the intermediate result of the view computation can be materialized in the warehouse. The incremental maintenance algorithm can implement warehouse real-time updating. We also discuss a strategy based on workflow to meet warehouse updating.The thesis also study the clustering application on transformer facility status analysis. A clustering analysis is made and the results show transformer facility status which can't be diagnosed by experience formula can be diagnosed accurately. |