NOx Concentration Prediction Model Based on Sparse Regularization Stochastic Configuration Network

Abstract

To accurately predict the nitrogen oxide (NOx) concentration during the municipal solid waste incineration (MSWI) process, a prediction modeling method based on sparse regularization stochastic configuration network is proposed. The method combines DropConnect regularization with L₁ regularization. On the basis of L₁ regularization constraint stochastic configuration network output weights, DropConnect regularization is applied to input weights to introduce sparsity. And a probability decay strategy based on network residuals is designed to address the problem of DropConnect fixed drop probability affecting model convergence. Finally, the generated sparse stochastic configuration network is used for NOx emission concentration prediction and experimentally validated using actual data from a solid waste incineration plant in Beijing. The results show that the proposed modeling method has high prediction accuracy and generalization ability, while effectively simplifying the model structure, which enables accurate prediction of NOx concentration.