关键词: 岩性识别, 径向基概率神经网络, 遗传算法

Abstract: Lithology identification is the most critical procedure in the logging data interpretation field, while the traditional lithology identification methods have a lot of defects such as slow explain efficiency, low accuracy, and big influenced human factors. To resolve these problems, a new kind lithology identification method was put forward using genetic optimized Radial Basis Probability Neural Network (RBPNN). Probabilistic Neural Network (PNN) and the Radial Basis Function Neural Network (RBFNN) were combined to construct RBPNN. To optimize network structure, upgrade convergence speed and accuracy, Genetic Algorithm (GA) was used to search for the optimal hidden center vector and matching kernel function control parameters of the RBPNN structure which must satisfy minimum error of RBPNN training and form genetic optimized RBPNN network model. The case study shows that lithology identification based on genetic optimized RBPNN can achieve the actual application standards, and it is feasible and effective, it also can provide scientific theoretical supports and dependences for oil geological exploration field.

Key words: lithology identification, Radial Basis Probabilistic Neural Network (RBPNN), Genetic Algorithm (GA