Anaerobic Digestion System Simulation Research Based On Data Mining Technology

Facing the energy shortage nowadays, making the various types of biomass waste into renewable energy with anaerobic digestion technology is of great significance to ensure sustainable energy supply in China.With the rapid development of information technology, the amount of data accumulated grows rapidly, then how to extract useful knowledge from vast amounts of data become a top priority, so data mining emerge as the times require.This article chose Artificial Neural Network(ANN) to do data mining, which based on the test data of two semi-continuous anaerobic co-digestion experiments with gradually increasing organic loading rate by using the kitchen waste with the pig manure, and the kitchen waste with landfill leachate respectively. VFA, pH, ammonia nitrogen, organic load, and former gas production were chosen as the control parameters of reaction process to establish gas production prediction model with Back Propagation Neural Network (BPNN) and Fuzzy Neural Network (FNN).About the experiment of kitchen waste and pig manure, the prediction accuracy was79.87%by the BPNN model and83.46%by the FNN model respectively. About the kitchen waste and landfill leachate, the prediction accuracy was81.92%by the BPNN model, and87.29%by the FNN model respectively.Comparing the results showed that the BPNN and the FNN both can be used to simulate anaerobic co-digestion system for gas production prediction. However, the stability of the test process will be a larger influence on the accuracy of the BPNN model, which the unstable stage will cause the worse accuracy. In contrast, the FNN model prediction accuracy had been greatly improved, even with shorter training time. It proved that joining fuzzy control into the traditional neural network could improve its accuracy, and be more suitable for the gas prediction of anaerobic co-digestion.