Building The Production Prediction Of The Horizontal Well Based On Data Mining Technology

The production predicting of the horizontal well makes an import effect on the process of horizontal well exploitation. The production predicting usually uses formulas at present, so the accuracy is low, and also the accumulated data for years do not be made well used of. The technology of data mining gives a better approach to acquiring information from vast data of horizontal well.The paper makes more study in the clustering analysis and genetic algorithm based on data mining technologies that existed. We proposed a genetic algorithm with a gene_center_orientation, which retained global optimization and solved the problem of fluctuation around the optimal results when genetic algorithm was running at the anaphase, and also the algorithm has enhanced the convergence rate. For the flaw that the clustering method plunged local optimization easily, we combined the GCOGA with hierarchical method, and used the global optimizion of genetic algorithm to make up the shortage of clustering method. When multiattribute swatches participated in clustering, we proposed a GCOGA-FCM which was combined with attribute of swatches. This new method completely considered the swatches attribute effect on the clustering effect.We have researched and improved the method of data mining, and applied it to the predicting production of the horizontal well. Aiming at the characteristic of predicting production of the horizontal well, we set up the model of the predicting production with the new arithmetic which was advanced in this paper. Experimental results showed that the model can accomplish thinning classes of horizontal well, and improve the accuracy of the predicting production. The application effect is comparative ideal.