支持向量回归法在抽油机系统效率信号经验模式分解中的应用  

Application of Support Vector Regression Method to Empirical Mode Decomposition Analysis of Oil Well System Efficiency

作　　者：原大宁[1] 张海峰[1] 刘宏昭[1] 饶建华[2] 

机构地区：[1]西安理工大学,西安710048 [2]中国地质大学,武汉430074

出　　处：《机械科学与技术》2008年第2期264-267,272,共5页Mechanical Science and Technology for Aerospace Engineering

基　　金：国家自然科学基金项目(50575180);陕西省科技攻关项目(2004K06-G23);陕西省教育厅科研项目(06JC07)资助

摘　　要：应用经验模式分解(EMD)方法分析抽油机系统效率变化的趋势项。讨论了EMD方法的端点效应。为了解决EMD的端点效应,首先应用支持向量回归(SVR)方法对抽油机系统效率测试数据进行回归和预测,与实测数据对比表明SVR回归和预测具有较高精度;然后应用SVR方法对系统效率测试原始数据进行双边延拓,对延拓后的数据信号进行经验模式分解。延拓前后分解所得的效率变化趋势对比表明,SVR方法可以有效地解决EMD的端点效应,提高抽油机系统效率变化趋势的预测精度。We apply the empirical mode decomposition （EMD） method to analyzing the trend of change in the efficiency of an oil well system and deal with the end effects of the EMD. First we employ the support vector regression （SVR） method to regress and predict the test data on the efficiency of the system. The comparison of the test data with measurement data shows that the regression and prediction with the SVR method are highly accurate. Then we carry out the bilateral extension of the crude test data of the systemts efficiency with the SVR method. The signal of bilaterally-extended data is decomposed via the EMD method. The comparison of the trend of efficiency change before and after extension indicate that the SVR method can effectively solve the end effects of EMD and enhance the accuracy of predicting the trend of change.

关 键 词：抽油机 系统 效率 支持向量回归 经验模式分解 双边延拓 

分 类 号：TH113[机械工程—机械设计及理论]