The Application of Improved BP Neural Network Algorithm in Lithology Recognition
Traditional technology of lithology identification bases on statistical theory, such as regression method and cluster method, which has some shortcomings. The standard BP neural network algorithm has some disadvantages like slow convergence speed, local minimum value which results in the loss of global optimal solution. BP neural network algorithm on the basis of improved variable rate of momentum factor can effectively overcome these disadvantages. Practical application shows that this method has the feature as high recognition precision and fast recognition rate so that it is suitable for recognition of lithology, lithofacies and sedimentary facies as well as geological research like deposit prediction and rock and mineral recognition.
KeywordsLithology Recognition Improved BP Algorithm Logging curves Momentum factor
Unable to display preview. Download preview PDF.
- 1.Cheng, G., Zhou, G., Wang, X.: The Probability Neural Networks for Lithology Identification. Control & Automation 23(6), 288–290 (2007)Google Scholar
- 2.Chen, R.: Research on the Application of BP Neural Networks. Control & Automation 23(8), 258–259 (2007)Google Scholar
- 3.Zhou, G., Cheng, G.: BRF neural networks for petroleum reservoir characterization. Computer Engineering and Applications 43(8), 174–176 (2007)Google Scholar
- 4.Zhou, B., Li, Z., Pan, B.: A Study on Lithology Identification Methods for Volcanic Rocks. Journal of Jilin University (Earth Science Edition) 35(3), 394–397 (2005)Google Scholar
- 5.Zhang, H., Zou, L., Shen, X.: The Application of BP Neural Network in Well Lithology Identification. Geology and Prospection 38(6), 63–65 (2002)Google Scholar
- 6.Xie, L.: BP Neural Network Algorithm Improvements and Convergence Analysis. Computing Technology and Automation 26(3), 52–56 (2007)Google Scholar