MODULAR ARTIFICIAL NEURAL-NETWORK FOR PREDICTION OF PETROPHYSICAL PROPERTIES FROM WELL LOG DATA

An application of Kohonen's self-organizing map (SOM), learning-vector quantization (LVQ) algorithms, and commonly used backpropagation neur al network (BPNN) to predict petrophysical properties obtained from we ll-log data are presented. A modular, artificial neural network (ANN) comprising a complex network made up from a number of subnetworks is i ntroduced. In this approach, the SOM algorithm is applied first to cla ssify the well-log data into a predefined number of classes. This give s an indication of the lithology in the well, The classes obtained fro m SOM are then appended back to the training input logs for the traini ng of supervised LVQ, After training, LVQ can be used to classify any unknown input logs. A set of BPNN that corresponds to different classe s is then trained. Once the network is trained, it is then used as the classification and prediction model for subsequent input data. Result s obtained from example studies using the proposed method have shown t o be fast and accurate as compared to a single BPNN network.