Study On Logging Data Lithology Identification Of Genetic-BP Neural Network Model In Bin-Chang Mining Area

The lithological interpretation is critical during logging data processing. The traditional manual interpretation has disadvantages of low efficiency, low precision and much useful information lost. With the development of digital geophysical logging technology, it has become the important aspect of the coal geological exploration to identify the lithology quickly and precisely and to improve the precision of geological interpretation for logging data using the advanced computer information process technology. In this paper, by analyzing international and domestic study on lithology identification, the widely applied Neural Network is selected as research emphasis. Based on the analysis of features of Genetic Algorithm (GA) and Back-Propagation Algorithm (BP), it is concluded that the disadvantage of BP algorithm includes large identification specimen count in inversion, slow constringency and easy to enter partial least status, so a methodogy to optimize BP network structure and link weights with GA is proposed and the lithology identification model based on GA optimized BP algorithm is established. In the last part of this paper, a software system for lithology identification, basic logging data management and column plotting is developed according to the model. Using the basic data from Bin-Chang mining area, the lithology identification function is tested, and the result indicates that the Genetic-BP Neural Network model has good identification speed and accuracy. Theresearch results of this paper are practical.