论文部分内容阅读
目前我国新发现的天然气田,有60%的天然气储量分布于低渗特低渗砂岩气田中,该类气田储量大,但储量风度低;产能低,产能变化大。由于低渗特低渗砂岩气层的产能受地层岩性、沉积微相、物性、孔喉结构、非均质性、油气充注和含气性等多种地质因素的复杂影响,导致产能预测与实际结果有较大的误差,严重影响后期的开发决策。广义回归神经网络(GRNN)本身具有很强的非线性映射能力,能逼近任意类型的函数。运用广义回归神经网络,综合地质、测井参数,并组合测井参数,敏感性分析,建立预测模型,对气层产能进行预测,取得相对较为满意的结果。
At present, 60% of natural gas reserves in China’s newly discovered natural gas fields are distributed in low-permeability and low-permeability sandstone gas fields, which have large reserves but low reserves, low productivity and large changes in production capacity. Due to the complex influence of many geological factors such as lithology, sedimentary microfacies, physical properties, pore-throat structure, heterogeneity, hydrocarbon filling and gas-bearing, the productivity of low-permeability and low-permeability sandstone gas reservoirs leads to the prediction of productivity With the actual results have a greater error, seriously affecting the development of late decision-making. Generalized Regression Neural Network (GRNN) itself has strong nonlinear mapping ability, can approach any type of function. Using generalized regression neural network, integrated geology, logging parameters, combined logging parameters, sensitivity analysis, the establishment of predictive models to predict gas production capacity, and get relatively satisfactory results.