基于模糊粗糙集的SVM污水处理过程故障诊断研究

he fault diagnosis on a waste water treatment process is a hard and complex problem. A fault diagnosis method was developed based on attribute selection with fuzzy-rough set (RSFS) and principle of support vector machine (SVM) classification. RSFS method is first used to reduce the redundant information and noise in the data. Meanwhile, SVM increases classification accuracy with good generalization performance. The resulting features are used to t rain a support vector classifier by solving a quadratic programming problem. Test results for fault detection in a real Wastewater Treatment Plan in Guangzhou show that the method has satisfactory accuracy.