Sludge bulking is a common and undesired phenomenon in wastewater treatment plants that negatively affects biomass settling characteristics, deteriorates treatment efficiency and causes severe operational problems. First-principles models for this phenomenon are not yet available. Therefore, data-driven models have been developed to predict sludge bulking. In this paper, the bulking phenomenon is studied from the control point of view and operating variables that can be used to control sludge bulking are identified. Identification is performed by designing a data-driven model using available process data as well as clustering and various classification methods. A global sensitivity analysis is applied to select the operating variables with the highest impact on sludge bulking. Application of the proposed approach to full-scale data has shown that increasing aeration intensity and limiting nitrogen sources are the most promising control actions for bulking control.

污泥膨胀是废水处理厂中常见且不受欢迎的现象，会对生物质沉降特性产生负面影响，降低处理效率并导致严重的操作问题。这种现象的第一性原理模型尚不可用。因此，已经开发了数据驱动模型来预测污泥膨胀。在本文中，从控制的角度研究了膨胀现象，并确定了可用于控制污泥膨胀的操作变量。识别是通过使用可用的过程数据以及聚类和各种分类方法设计数据驱动的模型来执行的。应用全局敏感性分析来选择对污泥膨胀影响最大的操作变量。