Determining Reservoir Properties in Reservoir Studies Using a Fuzzy Neural Network

Abstract

Abstract The uncertainty, fuzziness and incompleteness widely existent in reservoir studies require a powerful tool to deal with them more precisely. And fuzzy sets and neural networks seemed quite suitable for this purpose. In reservoir studies, it would be very advantageous to combine the symbolic level of processing conveyed by fuzzy sets with the computational power of neural networks. With the advantage of fuzzy sets in knowledge representation and the high capacity of neural networks in learning knowledge expressed in data, in this paper, a fuzzy neural network is used to solve for the inverse problems in reservoir studies with the purpose of determining the reservoir properties from well logs. This paper presents a fuzzy neural network approach for determining reservoir properties from well logs, based on fuzzy sets and a self-organizing mapping neural network. Both the details of the algorithm of this fuzzy neural network and its implementation are given. A field example for recognizing lithology from well logs is used to demonstrate the dependability of this fuzzy neural network approach.