A Digester Temperature Prediction Model Based on the Elman Neural Network

Abstract

<abstract> Anaerobic digestion can proceed stably and efficiently within an appropriate temperature range, and sharp temperature fluctuations could lead to a reduction in biogas production. Therefore, the ability to predict temperature is important for the anaerobic digestion process. In this study, the Elman neural network (Elman NN) was developed to simulate and predict temperature changes in the digestion process. The simulation results showed that the model was able to fairly accurately predict short-term temperature changes, which was evidenced by the fact that (1) the absolute errors between the temperatures predicted by the Elman neural network model and those measured using the Toky-100 sensor were less than 0.5°C in most cases and (2) the temperature predicted by the Elman neural network model were consistent with those measured by the Toky-100 sensor under different weather conditions and at different locations. Predicted temperatures are useful for optimizing digester operation, overcoming temperature hysteresis, and intelligently diagnosing potential malfunctions of the digestion process.