基于改进差别矩阵属性约简的聚合釜粗糙集-神经网络故障诊断

引言聚氯乙烯树脂(PVC)是重要的有机合成材料,其产品具有良好的物理性能和化学性能,广泛应用于工业、建筑、农业、电力、公用事业等领域.聚合釜则是聚氯乙烯生产装置的关键设备,聚合釜能否稳定运行直接关系到整个聚氯乙烯生产装置的运行状况.

Rough set-neural network fault diagnosis of polymerization based on improved attribute reduction algorithm of discernibility matrix

Aiming at the real-time fault diagnosis and optimized monitoring requirements of the polymerizer of PVC production process, a real-time polymerizer fault diagnosis strategy was proposed based on rough set (RS)theory with improved discernibility matrix and back propagation (BP)neural network.The improved discernibility matrix was adopted to reduce the attributes of rough set in order to reduce the input dimensionality of fault characteristics effectively.Fuzzy C-means clustering algorithm was used to discrete the continuous variables of the decision table.Then Levenberg-Marquardt BP neural network was trained according to the reduced decision table in order to decide the configuration parameters of the proposed polymerizer fault diagnosis model.Thus the classification of the fault patterns was to realize the nonlinear mapping from fault symptom set to polymerizer fault set according to a set of symptoms.Polymerizer fault diagnosis simulation experiments were performed by combining with industry history data.Simulation results showed the effectiveness of the proposed fault diagnosis method based on rough set and BP neural network.