Modular artificial neural network for prediction of petrophysical properties from well log data

This paper reports the application of Kohonen's Self-Organizing Map (SOM) and Learning Vector Quantization (LVQ) algorithms, and the commonly used Back Propagation Neural Network (BPNN) to the prediction of petrophysical properties from well log data. Recently, the use of artificial neural networks (ANN) in the field of petrophysical properties prediction has received increasing attentions. In this paper, a modular ANN comprises of a complex network made up of a number of sub-networks is introduced. In this approach, the SOM algorithm is first applied to classify the well log data into a pre-defined number of classes. This gives an indication of the lithology of the given well. The LVQ algorithm is then applied to train the network under supervised learning. A set of BPNN which corresponds to different classes is then developed for the prediction of petrophysical properties. Once the network is trained if is then used as the classification and prediction model for subsequent input data. Results obtained from example studies using this proposed method have shown to be fast and accurate as compared to a single BPNN network.