Nox浓度的混合正则化块增量随机配置网络预测方法OA

To achieve rapid and accurate prediction of NOx emissions in Municipal Solid Waste Incineration(MSWI)processes,this paper proposes a method based on a Mixed Regularization Block incremental Stochastic Configuration Network(MR-BSCN).After completing the model training with BSCN,redundant nodes are pruned by incorporating a momentum term into the convex optimization-approximated L0 regularization.Meanwhile,to ensure the accuracy of the pruned model,L2 regularization is applied to fine-tune the output weights.Finally,verification using real-world data from an MSWI plant in Beijing demonstrates that the proposed method achieves accurate NOx concentration prediction with a more compact model structure,building upon the BSCN's fast modeling capability.This work lays a foundation for the optimal control of NOx emissions.