Real-time monitoring of decentralised anaerobic digestion using artificial intelligence-based framework

This paper presents an Artificial Intelligence (AI)-based framework for real-time monitoring and improving the operation of an Anaerobic Digestion (AD) system in producing biogas. This was achieved using historic data obtained from a decentralised AD plant located in Camley-Central London to develop a recurrent neural network (RNN) model based on AI to predict biogas production with respect to lag time. The dataset obtained from the AD plant had a wide range of missing values, which hindered the accurate prediction of biogas. This study evaluates different data mining techniques for infilling missing data. The Recurrent Neural Network (RNN) Model was then developed for predicting biogas with respect to various lag times. The results show both Kriging and Linear Regression techniques have the best performance, and they were used to infill the missing data. The results also show biogas production can be accurately predicted in real-time operation using a NARX model based on the feed data including organic food composition such as oats, soaked liners, catering and water added.