用正交尺度小波网络方法预测固井质量  

Cementation Quality Prediction Using Wavelet Neural Network Based on Orthogonal Scaling Function

作　　者：艾池[1] 卜志丹[1] 赵万春[1] 李强[2] 

机构地区：[1]大庆石油学院提高油气采收率教育部重点实验室,黑龙江大庆163318 [2]大庆油田有限责任公司第九采油厂,黑龙江大庆163318

出　　处：《石油钻探技术》2008年第6期56-58,共3页Petroleum Drilling Techniques

摘　　要：分析了固井质量预测系统的复杂性和正交尺度小波网络的优点,采用SAS系统对影响固井质量的众多因素进行了相关分析,通过正交尺度小波网络建立了固井质量预测模型。该模型以影响固井质量的主要因素地层压力系数、渗透率、井眼扩大率、井眼规则度、钻井液密度、水泥浆密度、套管居中度和顶替返速作为预测模型的输入参数,将固井质量定量化作为模型的输出。预测结果与实际检测结果的最大相对误差为6.87%,且计算速度快,大大节省时间,因此该模型具有较好的应用前景。The advantages of orthogonal sealing function wavelet neural network and the complexity of cementation quality forecasting system were analyzed. Factors affecting cementing quality were analyzed using SAS system. A cementing quality prediction model was developed using wavelet neural network based on orthogonal scaling function considering the main factors. Input data of this model include main factors affecting cementing quality, including formation pore pressure, permeability, wellbore enlargement, wellbore diameter, drilling fluid density, cement slurry density, casing eccentricity, and displacement velocity. Cementing quality is output as a quantified number. The maximum relative error between prediction result and actual result is 6.87%, the calculation is fast which saves time. Therefore, the new model has a good future in application.

关 键 词：固井质量 小波神经网络 预测 尺度函数 

分 类 号：TE928[石油与天然气工程—石油机械设备]