摘要
地球化学方法在天然气水合物勘探评价过程中的参数存在不确定性,且误差传递易导致结果不可信。运用BP神经网络技术,在天然气水合物勘探区域选取相关的应用切入点,通过训练建立神经网络模型,利用其非线性映射技术,揭示天然气水合物勘探评价中涉及的多个属性之间的非线性关系。计算结果显示,神经网络的分类方案有效弥补了当前地球化学评价方法存在的多解性等缺点,运用在地球化学数据的基础上建立的BP神经网络模型,对研究区块进行仿真预测,可以实现水合物矿藏的分等级评价。
Geochemical methods used in the exploration of natural gas hydrates have uncertainty of parameters, and are in lack of result credibility due to the error transfer. This study applied the artificial neural network technology as a breakthrough point to explore natural gas hydrates, and established a neural network model through training. Using the nonlinear mapping technique, we revealed the nonlinear relationship among the multiple attributes during the evaluation of natural gas hydrates. The calculation suggests that the clas- sification of the neural network can effectively remedy the defect of multiple solutions. It is illustrated that the BP neural network model based on geochemical data can simulate the study area and can further realize the classified evaluation of natural gas hydrates.
出处
《地质学刊》
CAS
2016年第1期113-117,共5页
Journal of Geology
基金
中国地质调查局项目"天然气水合物专项数据库建设及战略研究"(GZH201100312)
关键词
地球化学分析
BP神经网络
网络训练
网络仿真
geochemical analysis
BP neural network
network training
network simulation