Fault Diagnosis of Imperial Smelting Furnace Based on Rough Sets and Neural Networks
Hongqiu Zhu · Control Engineering of China · 2008
Considering the complexity of mechanism and uncertainty of information in the imperial smelting process,a fault diagnosis method based on rough set(RS)theory and neural network is discussed.Continuous attributes are discretized by combining SOM and the dependence of attributes.An advanced attribute-importance computing method is presented based on expert experiences and dependence of attributes,and is used in the heuristic reduction of RS to reduce the samples.The simulation results show that the structure of neural network is optimized,the computation complexity is decreased,and the diagnosis correctness are greatly improved after the training data is processed by RS.