Robust committee machine for water saturation prediction

• We introduced an intelligent structure named RCM for water saturation prediction.
• This framework provides high accuracy, generalization, fast and low cost for our object.
• Proposed CM include 16 experts based on NNs, FLs and NFs.
• We proposed a pruning method based on GA to find out an optimum sub-CM.
• We have used Fuzzy Genetic Algorithm as combining method for constructed sub-CM.

Water saturation is one of the important physical properties of the petroleum reservoir which are usually determined by core analysis. An accurate determination of this parameter is significant to execute a realistic evaluation of hydrocarbon reserves in the formation and also decreasing the economic risk. In this study, a robust technique is proposed to determine an accurate value of this parameter from well log data in un-cored well or at un-cored interval of the same well by combining different types of machine learning techniques. The final results (sub-CM outputs) demonstrated that integrating these techniques using proposed method provides an accurate, fast and cost-effective method for estimating the target value.