Vol.52 (AICT 2014), pp.60-64

The regional engineering geological conditions and geological hazard evaluation is the important part of realizing the railway geological selection, However, the accuracy of the evaluation largely depends on the reliability of basic information and reasonable mathematical model. Remote sensing provides reliable geological information and accurately reflects the geological conditions along the railway line. The artificial neural network has a self-learning ability, even with the noise data lack of items or lack of awareness still can get a satisfactory conclusion. Therefore, the paper uses the improved BP neural network-LM neural network established evaluation and prediction model, Achieved the effective assessment and prediction of engineering geological conditions and geological hazard align the railway. Theoretical analysis and case study shows, the Levenberg-Marquard neural network based on the adaptive adjustment learning step has high accuracy and speed advantages, Is an ideal geological hazard assessment methods.