Design of a prediction model based on improved BP neural network and particle swarm optimization for more accurate budget of biogas production

Abstract

In order to accurately predict the daily gas production of large and medium-sized biogas projects, the improved BP neural network algorithm was used, and the PSO algorithm was introduced to optimize the parameters. According to the anaerobic fermentation mechanism and the actual engineering operation status, a prediction model was established with temperature, daily feed volume, NH3, TS concentration and pH value as input layer nodes, and daily biogas production as output layer nodes. The 116 sets of data obtained by remote data acquisition are used as training samples and test samples of the model, and the simulation is carried out by Matlab software. The results show that the PSO-LM-BP neural network has good predictive ability for the daily gas production of biogas. The established biogas daily gas production prediction model not only converges fast but also has high accuracy.