数据挖掘技术在出砂预测中的应用

Sanding Prediction : Data Mining Approach

针对油井出砂预测困难的问题 ,提出应用数据挖掘理论和技术探索解决问题的新途径。根据应用问题的性质 ,给出了两种用于实现聚类挖掘的人工神经网络模型与算法 ,简单竞争学习和自组织特征映射学习算法。运用该挖掘模型对某油田一区块的 5 2口已知出砂效果井的资料进行挖掘 ,预测该区另外 8口井出砂情况 ,在预测结果中有 5口井的出砂预测结果与实际出砂情况相符。

Data Mining theory is introduced to predict the sanding which is difficult to solve with traditional methods. Two artificial neural network algorithms, simple competitive algorithm and solf-organized mapping algorithm are presented in this paper.They are suitable for clustering data mining. An instance is provided to demonstrate the application of data mining. It shows that five out of eight is correct in the application.