Research Of Multidimensional Data Modeling And Materialized Views In Load Forecasting

Based on the characteristics of the historical data set of load forecasting system, forecasting methods and the requirements of OLAP and DM tools for data sources, this paper researches the theories of multidimensional data model and materialized views maintenance in order to improve the independence and maintainability of the data sources and the efficiency of the OLAP and DM.The paper has four chapters. In the first chapter, the importance, present situation and problems of load forecasting are introduced simply. To solve the problems, a strategy of building a data warehouse to support the load forecasting system is proposed. To build the data warehouse, there are two important theoretical problems about multidimensional data modeling and materialized views. Both the problems are emphasized at the end of the chapter, and discussed profoundly in the following chapters.The second chapter begins with the general review of the importance and researching status of multidimensional data modeling, then points out the limitations of the existing multidimensional data models. In order to model the multidimensional structure of the load forecasting system's data set, at the end of the chapter, the paper introduces a new model to solve the problems of different dimensions sharing a common bottom level of hierarchies and incorporating incomplete data set of fine granularity and complete data set of coarse granularity logically to support OLAP which had not been solved by prevenient models.First, the third chapter talks about the researching achievements on materialized views, and then on the basis of load forecasting, presents a. new materialized maintenance algorithm, which can reduce the space and time cost of the materialized views maintenance.The fourth chapter discusses how to support the load forecasting using the multidimensional data model and the view maintenance algorithms that the paper presents before.Finally, the conclusion of the paper is given and the future work is suggested.