KPCA and ELM ensemble modeling of wastewater effluent quality indices

Abstract

Reliable measurements of effluent quality are important for different operational tasks such as process monitoring, online simulation, and advanced control in the wastewater treatment process (WWTP). A kernel principal component analysis (KPCA) and extreme learning machine (ELM) based ensemble soft sensing model for effluent quality prediction was proposed. KPCA was used to extract nonlinear feature of input space to overcome high dimension and colinearity. ELM algorithm is inserted into the ensemble frame as a component model since ELM runs much faster and provides better generalization performance than the other popular learning algorithm. The average output of all the ELM components in the ensemble is the final estimation of the effluent quality index. Simulations results using industrial process data show that the reliability and accuracy based KPCA and ELM ensemble soft sensing outperform the ELM, ELM ensemble model.