L-M neural network for data mining of oil saturation

摘要

Basing on the geological data base, the oil saturation of the reservoir is evaluated by data mining and knowledge discovering with artificial neural network. To get better convergence, faster convergent speed and higher precision of the neural network, Levenberg-Marquardt algorithm is adopted to improve the learning algorithm of the neural network, which leads to global convergence with faster convergent speed than BP algorithm. The principle and the procedures of oil saturation data mining as well as knowledge discovering with L-M neural network are then discussed. An engineering case is also presented to explain and testify the method proposed.