Research on prediction method of sludge bulking based on ANN and grey Markov model

Sludge volume index (SVI) can evaluate and reflect the aggregation of activated sludge sediment properties accurately. It is an important parameter to predict sludge bulking. Generally, if SVI value is too high, the description is sludge settling performance is poor. It will occur or has occurred sludge bulking. But SVI cannot be online measurement, offline assay data obtained for a long time or other issues. To solve this problem, this paper has applied soft-sensing technology for the sludge volume index that reflects sludge bulking, using rough set to reduce the instrumental variables then construct the soft-sensing model with RBF neural network to complete the dataset of sludge volume index, and then, employed the grey Markov model to predict the dataset to collect the important information of sludge bulking in the quantitative respect, in order to achieve real-time prediction of sludge bulking.