论文部分内容阅读
在BP(ErrorBack Propagation)算法基础上 ,采用BI(BackImpedance)算法 ,运用了自组织优化隐层节点数和自动优化网络因子的方法 ,使得人工神经网络ANN的计算速度、精度和柔韧性有所提高 ,且在微机上其操作变得更加容易 .在勘探成熟的气藏中 ,按照天然气成藏理论 ,选取能够系统反映气藏的 8个储量评估参数 ,进行网络学习 ,建立储量评估模型 .应用所建的网络模型对正处于勘探阶段的气藏进行了很好的储量等级识别 ,同时 ,利用改进后网络精度高的特点 ,尝试着进行了储量估算 ,获得较好的效果 .这种改进的ANN为天然气勘探提供了一种新的储量评估方法 .
Based on BP (ErrorBack Propagation) algorithm, BI (BackImpedance) algorithm is used to optimize ANN speed, precision and flexibility by using self-organized optimization of hidden layer nodes and automatic optimization of network factors , And its operation on the computer becomes easier.According to the theory of natural gas accumulation in the mature gas reservoirs, 8 evaluation parameters that can systematically reflect the gas reservoirs are selected and network learning is conducted to establish a reserve evaluation model. The built network model of gas reservoirs is in the exploration stage has a good reservoir level identification, at the same time, the use of improved network features high accuracy, try to estimate the reserves, and get better results. The improved ANN A new method of reserves evaluation is provided for gas exploration.