Font Size: a A A

Reservoir Prediction Method Of Nonlinear Analysis

Posted on:2006-12-03Degree:MasterType:Thesis
Country:ChinaCandidate:W J WangFull Text:PDF
GTID:2190360155458443Subject:Applied Mathematics
Abstract/Summary:PDF Full Text Request
In recent years, it is widely researched on the study algorithm of the neural network with multi-layer forward. Thus, the principle of the neural network with multi-layer forward was deeply studied in this paper, and some research evolutions which had already obtained were summarized. Also, the author suggested Damped Least Squares method and Particulate Swarm Optimization, in allusion to the question and the limitation exists in the BP algorithm based on the Steepest Descent Method, which made application for the forecast of the layer parameters and the lithology recognition, at the same time it improved the iterative speed and precision. At the same time, it put forward an algorithm for optimization of neural network architecture with high efficiency.The whole process and the key technology which the neural network used for logging well and storing deduce were introduced in detail in this paper. That is, how to choose the study stylebook; how to optimize the structure of the network; how to improve the ability of the extension of the neural network; how to do with the data ofthe logging well and so on.In order to improve the veracity of the layer and the gas and improve the application technology in the intelligent level and adaptation. It is to perform instead, compared BP neural network law and can be seen with routine mathematics and physics precision of statistical method layers of characteristic to apply to, neural network parameter of law predict precision have greater improvement, demonstrate in store layers of parameter predict advantage of and use latent energy. On the basic of the gas structure in Hardson oil gas field in Xingjiang Province, Qiaokou oil gas field in Shandong Province, Luodai oil gas field in Sichuan Province, dealt with nearly hundred of logging well data in complex geological conditions. By the computation methods of the neural network, the reservoir characters(such as porosity, permeability) was forecasted and organized, also the precision was tested, which obtained a obvious geography efficiency superior to traditional explanation methods.It is shown according to the result that the computation method of the neural network not only can conquer the trouble in the complicated nonlinear modeling in the general logging well, also it predigest the mathematical methods in the explanation.
Keywords/Search Tags:artificial neural networks, porosity, permeability, lithology recognition, intelligent inversion
PDF Full Text Request
Related items