Prediction of Temperature Conditions of Autothermal Thermophilic Aerobic Digestion Bioreactors at Wastewater Treatment Plants

Abstract

Energy integration plays a significant role in increasing energy efficiency and sustainability of production systems. In order to model real energy integrated systems, sometimes we don't need rigorous models for involved units, but easily implemented and fast ones instead. This study presents an approach based on Artificial Neural Networks (ANNs) for predicting the main parameters of industrial Autothermal Thermophilic Aerobic Digestion (ATAD) bioreactors that are crucial for their energy integration. To create such predictive ANN model, four architectures with different number of hidden layers and artificial neurons in each one of them have been investigated. The developed ANN architectures have been trained and validated with data samplings obtained through long-term measurements of the operational conditions of real ATAD bioreactors. To train the models, BASIC genetic algorithm has been implemented. Using three independent measures for validation of the models, the best ANN architectures were selected. It is shown that selected ANN models predict with sufficient accuracy these ATAD parameters and are suitable for the implementation in an energy integration framework.

Keywords

• Wastewater treatment plant
• Autothermal thermophilic aerobic digestion bioreactor
• Parameters prediction
• Artificial neural network
• Genetic algorithm