Dissertation > Excellent graduate degree dissertation topics show

# Data-driven Subspace Methods and Its Application in the Prediction

Author: SunCe
Tutor: GaoZhen
School: Northeastern University
Course: Systems Engineering
Keywords: Data-driven Subspace Particle swarm optimization Prediction ofreservoir parameters Energy consumption
CLC: N945
Type: Master's thesis
Year: 2011
Quote: 0

### Abstract

 Energy is the foundation of modern industry. Oil and natural gas, as the important energy resources and strategic resources, widely affect the economic development of all countries. On the other hand, iron&steel industry, as one of the most important basic industries of our country, is a high energy consumption industry. Therefore, oil and natural gas are the material basis of iron&steel industry and at the same time, iron&steel industry can provide all sorts of equipments which can promote the development of the oil&nature gas industry. Using scientific technological methods to strengthen the predictions of the oil reservoir and energy consumption in the iron&steel industry could not only relieve the energy tension of our country, but also can promote the development of the iron&steel industry and the oil&gas industry rapidly.Since been proposed in1990s, subspace method has been successfully used for system identification and predictive control and other fields. With the development of technology, the quantity of data generation and storage increases and the data-driven subspace method was proposed. The prediction model of data-driven subspace algorithm is selected to apply in the energy prediction problems in this work. To solve the above two problems, the major work of this paper are as follows:(1) Data-driven subspace algorithm and the improved one. Use the prediction model of the data-driven subspace algorithm to predict two issues. In order to improve the precision of prediction, the two factors are introduced and the PSO algorithm is used to optimize the two factors.(2) The prediction of reservoir parameters. Reservoir modeling is the core task of reservoir enterprise description. The prediction of reservoir parameters is the foundation of the reservoir modeling. From the spatial dimension, this paper uses data-driven subspace method to predict the main reservoir parameters. The introduced factor can improve parts of the results and the dissertation analyzes the reasons.(3) The prediction of energy consumption in iron&steel enterprise. From the time dimension, this work uses data-driven subspace methods to predict the energy consumption for real production of some iron&steel enterprise. The improved algorithm can improve the precision of prediction.(4) Development of energy forecasting system for iron&steel enterprise. The embedded algorithm can provide the predicted results of the consumption of energy quickly and accurately. The system can display the results in different ways, such as tables; graphs, bar charts, pie charts and so on. In addition, the aggregation function and manual adjustment functions are also incorporated.

### Related Dissertations

CLC: > SCIENCE AND > Journal of Systems Science > Systems Engineering
© 2012 www.DissertationTopic.Net  Mobile