Font Size: a A A

Reservoir Parameter Modeling Research Based On Neural Network

Posted on:2007-07-25Degree:MasterType:Thesis
Country:ChinaCandidate:Y R YangFull Text:PDF
GTID:2120360185966973Subject:Control theory and control engineering
Abstract/Summary:PDF Full Text Request
Oil reservoir parameter has sorts of complicated influent factors, and many objectives are uncertainty and change with time. This brings some difficulty to modeling and improving the accurate of forecasting modeling. So the requirement to the method of reservoir parameter prediction is more and more high. It has an important meaning to improve extraction ratio of oil field and prolong field developed service life, which adopt an operable technique method and improve the coincidence rate of reservoir parameter prediction.Basing on the mentioned objective above, this paper have did several aspect of work as follows.(1) Analyze the technique characteristics of kinds of oil parameter estimate methods in petroleum explore, and the shortage itself in solving the geology problems;(2) Analyze some problems and difficulties of BP algorithm may meet when used in practice, discussed the reason that generate these problems and solve method;(3) Analyze and discuss the long train time of neural network may meet in practice, the difficulty modeling and other problems, put forward a kind of practical method;(4) Study and compare some methods of building year model, compare each relative merit of RBF network and BP network, Dig into the particle swarm optimization algorithm, the performance comparing of particle swarm, studying the training algorithm that particle swarm combines with the neural network, discussing some problem for particle swarm to use for the neural network training. Aim at deal with the great data in establishing year model; research the method of grouping data to modeling;(5) In order to observe the validity of method that the paper puts forward, the reservoir parameter forecasting system is developed with the visual technology.
Keywords/Search Tags:neural network, forecasting, particle swarm, optimization algorithm
PDF Full Text Request
Related items