Characteristic Variable Selection Method and Predictive Modeling for Municipal Solid Waste Heat Value

In the process of municipal solid waste incineration, the fluctuation of waste heat value affects the stability of waste incineration. To make the real-time online prediction and change the trend of the waste heat value, the fuzzy neural network soft sensing method was adopted, and the on-line operation data of the incineration power plant was used as the input to accomplish the real-time prediction function of the waste heat value. First, the mutual information method was used to eliminate irrelevant variables from characteristic variables. Then, the fuzzy neural network and particle swarm optimization algorithm were combined to further eliminate redundant variables from the selected characteristic variables, so as to determine the input variables for predicting the waste heat value, and the fuzzy neural network prediction model for waste heat value was trained. Finally, the performance test was carried out through the collected sample data. Results show that this method has good prediction accuracy and real-time performance, and is suitable for online prediction of waste heat value.