基于SVM的高含水率油井常温集输可行性分析及能耗优化

Feasibility analysis and energy consumption optimization of high water bearing oil well gathering and transportation at normal temperature based on SVM

为降低高含水率油井在集输过程中的热力能耗,实现生产效益的有效提升,利用目前已实现常温输送的油井数据,通过熵权-灰色关联算法确定影响高含水率油井常温输送的主要因素,建立基于支持向量机(SVM)算法的油井常温集输可行性预测模型,并根据预测结果进行能耗分析。结果显示,埋地温度、产液量、含水率、气液比和集油管长等因素对实现高含水率油井常温集输的影响较大;当隐含层节点数为19时,SVM模型的预测效果最好,实验区块上的分类准确率为96.66%,模型具有较好的泛化性能;高含水率油井实现常温输送后,年可节约药剂、热洗和燃料等费用,合计4319万元,节能潜力巨大。该模型可在其他区块或油田推广使用。

In order to reduce the thermal energy consumption in the gathering and transportation process of high water bearing wells,the production efficiency can be effectively improved.Using data from wells where normal temperature transport has now been achieved,the main factors affecting the normal temperature transportation of high water bearing wells are determined by entropy weight and grey correlation algorithm,and the feasibility prediction model of normal temperature gathering and transportation of oil wells is established based on support vector machine(SVM)algorithm.The energy consumption is analyzed according to the predicted results.The results show that buried temperature,liquid yield,water bearing rate,gas-liquid ratio and collecting pipe length have great influence on the collection and transportation of high water bearing wells at normal temperature.When the number of hidden layer nodes is 19,the prediction effect of SVM model is the best.The classification accuracy of experimental block is 96.66%,and the model has good generalization performance.After realizing normal temperature transportation in high water bearing wells,the cost of agent,hot wash and fuel can be saved,totaling 43.19 million yuan,which has great energy conservation potential.Even more to the point,the model can be applied to other blocks or oilfields.