Journal of Petroleum & Environmental Biotechnology

A method for log interpretation based on modified fuzzy neural network

4^th International Conference on Petroleum Engineering

August 15-17, 2016 London, UK

Xiao-hui Zeng and Ya-juan Xue

Chengdu University of Information and Technology, China

Posters & Accepted Abstracts: J Pet Environ Biotechnol

Abstract:

With the development of oil exploration, there is more and more ambiguity in the conventional logging interpretation. Although complex reservoirs and oil content identification can be analyzed and predicted by means of traditional neural network, there are some defects in information amalgamation especially when resolving different nonlinear problems. So we design a modified self-organizing neural network algorithm for qualitative attributes reduction integrated with rough set. Firstly log data with some qualitative attributes are analyzed in an information system from a view of fuzzy set, and then the key attributes from the interacted attributes with oil-bearing formation are extracted. Secondly unscented particle filtering(UPF) are used in estimating the parameters of the self-organizing fuzzy neural network, finally the experiments demonstrate that our algorithm can solve the predicting problems of nonlinear system with constraints, and extracts the if-then fuzzy rules in oil field with the less attributes and higher accuracy. The approach also shows rapid computing speed and strong anti-disturbance capacity. It will be verified to be suitable for lithology recognition and oil log interpretation in actual environments.

Biography :

Email: huizi003@126.com