鱼群算法优化组合核函数GPR的油井动液面预测

发布时间：2018-01-10 10:30

本文关键词：鱼群算法优化组合核函数GPR的油井动液面预测　出处：《东北大学学报(自然科学版)》2017年01期 　论文类型：期刊论文

【摘要】：针对抽油井动液面(DFL)检测主要依靠人工操作回声仪测试,无法实时在线检测,而单一核函数的高斯过程回归(GPR)无法明显提高预测精度和泛化能力,提出了一种人工鱼群算法(AFSA)优化组合核函数的动态高斯过程回归动液面预测模型.采用多项式函数、线性函数与径向基函数组合构建核函数,利用人工鱼群算法对核函数模型参数进行寻优,采用快速傅里叶变换(FFT)和核主元分析(KPCA)融合提取时频数据非线性特征作为模型输入,提高模型的预测精度和泛化能力.油田现场应用验证了该方法的有效性.
[Abstract]:In view of pumping well moving liquid level DFL detection mainly depends on manual operation echo testing, can not be real-time on-line detection. But the single kernel function, Gao Si process regression (GPRR), can not obviously improve the prediction accuracy and generalization ability. In this paper, an artificial fish swarm algorithm (AFSA) is proposed to optimize the dynamic Gao Si process regression model for predicting the level of moving liquid surface. The kernel function is constructed by combining polynomial function, linear function and radial basis function. Artificial fish swarm algorithm is used to optimize the parameters of kernel function model. The fast Fourier transform (FFT) and kernel principal component analysis (KPCA) fusion are used to extract the nonlinear features of time-frequency data as the input of the model. The prediction accuracy and generalization ability of the model are improved.
【作者单位】：东北大学信息科学与工程学院;渤海大学工学院;
【基金】：国家自然科学基金资助项目(61573088,61433004,61403040)
【分类号】：TE33;TP18
【正文快照】： 游梁式抽油机井有杆泵采油是国内外油田普遍使用的机械采油方式[1-5].抽油井动液面(dynamic fluid level,DFL)是指油井正常生产时,油、套管环型空间内的液面,它受油层压力和采油速率的约束,是反映油层供液能力和井下供排关系的重要工况参数.文献[6-8]采用高斯过程回归(Gaussia

【相似文献】