Hierarchical Neural Network Modeling Approach to Predict Sludge Volume Index of Wastewater Treatment Process

From a practical-theoretic viewpoint, there is a need to develop rigorous design and analysis tools for the control, fault diagnosis, and security of wastewater quality. However, sludge bulking remains a widespread problem in the operation of activated sludge processes, which leads to severe economic and environmental consequences. Sludge volume index (SVI) monitoring is a key challenge that will become even more crucial in the years ahead to quantify sludge bulking. This brief presents a system that consists of online sensors and an SVI predicting plant. The SVI predicting plant uses a hierarchical radial basis function (HRBF) neural network to predict SVI in a wastewater treatment process (WWTP). Then, an approach named extended extreme learning machine (EELM) is proposed for training the weights of HRBF. Unlike conventional single-hidden-layer feedforward networks, this EELM-HRBF is based on the hierarchical structure which is capable of hierarchical learning of sequential information online, and one may only need to adjust the output weights linking the hidden and the output layers. In such EELM-HRBF implementations, the EELM provides better generalization performance during the learning process. Moreover, the convergence of the proposed algorithm is analyzed. To illustrate the methodology, the proposed predicting plant with the EELM-HRBF has been tested and compared with other methods by applying it to the problem of predicting SVI in a simplified and real WWTP. Experimental results show that the EELM-HRBF can be used to predict the wastewater quality online. The results demonstrate its effectiveness.