采用支持向量机算法优化电化学处理油田污水的工艺参数

摘要

采用支持向量机(SVM)算法,将Box-Behnken设计法与支持向量回归算法(SVR)实验参数优化软件相结合,优化电化学去除油田污水COD的工艺参数.通过量子粒子群算法对SVM算法参数进行优化,从建立的回归模型中找到工艺参数的全局最佳点:电解时间60 min,电解电流3A,三维电极填充料中石英砂质量695 g.模型得到的COD理论最优去除率为92.48％,验证实验得到的COD去除率为91.43％.%Combining the Box-Behnken design method with support vector regression algorithm (SVR,a software for experimental parameter optimization),the process parameters for electrochemical removal of oilfield wastewater COD were optimized using the support vector machine (SVM) algorithm.The SVM algorithm parameters were optimized by quantum particle swarm algorithm,the global optimal point of the parameters were found out from the regression model,such as:electrolysis time 60 min,eleOrolytic current 3 A,mass of quartz sand in the three dimensional electrode filler 695 g.The optimal COD removal rate from the model was 92.48％,while that from the experiment was 91.43％.