Abstract: A fast and practical backpropagation algorithm--resilient backpropagation (RPROP) has been introduced to better solve lithologic identification problems using well logging data. A backpropagation neural network model of lithologic identification based on the RPROP algorithm is established to study a real well logging data. The results indicate that the accuracy of identification is high and the RPROP algorithm is fast and practical compared with conventional backpropagation algorithm and some other modified backpropagation algorithm.

Key words: resilient backpropagation(RPROP) algorithm, backpropagation neural network, logging data, lithologic identification