基于输入参数优化选择的BP神经网络预测垃圾焚烧酸性气体原始排放浓度研究

Study of BP Neural Network for Predicting Original Emission Concentration of Acid Gas of Waste Incineration Based on Optimal Selection of Input Parameters

为准确预测垃圾焚烧发电厂SO_(2)、HCl酸性气体原始排放浓度,采用Copula函数探讨了多运行参数与酸性气体排放之间的相关性大小,并选择相关性较大的参数作为原始浓度预测的输入参数,建立了酸性气体原始浓度BP神经网络预测模型。以某垃圾焚烧电厂实测运行数据进行实例分析,证明了上述分析方法和预测模型的有效性。

In order to accurately predict the original emission concentrations of SO_(2) and HCl acid gases from waste-to-energy incineration plants,a BP neural network prediction model for original concentrations of acid gas was established by using the Copula function to explore the magnitude of correlation between multiple operating parameters and acid gas emissions,and the parameters with greater correlation as input parameters for original concentration prediction were selected.The effectiveness of the above analysis method and prediction model is demonstrated by an example analysis with actual measured operational data of a waste incineration plant.