相同构造区新型产量预测方法研究

Research on A New Production Prediction Method in the Same Tectonic Regions

基于主成分分析与神经网络方法,以焦页AHF井、BHF井为例,在不减少变量信息的情况下利用降维的思路将多个影响变量转化为少数综合变量,通过神经网络气井产量预测模型对输入集进行反复训练,得到输出变量。实验结果表明:该气井产量预测模型一致性较好,具有较高的精度,可以利用此方法进行初期产量预测。

On the basis of principal component analysis and neural network method and without reducing variable information, the idea of dimensionality reduction is used in Jiaoye AHF well and BHF well to translate multiple effect variables into a few aggregative variables. The output variable is obtained through repeated input set training based on neural network gas well production prediction model. Experimental results show that the model has a good consistency and high precision. Such a method can be used in initial production prediction.