Study On Slagging Of Biomass Combustion Based On Spectral Characteristics

Biomass fuel has the characteristics of low carbon and cleanness.It is the main component of renewable energy,and it has broad development and application prospects.Biomass fuel is rich in metal elements,which will produce slagging during the combustion.It not only affects the heat transfer efficiency,but also affects the safe operation of the boiler.The combustion flame reflects the combustion conditions,especially the spectral information of the alkali metals K and Na in the flame,which is closely related to the biomass fuel slagging.In this paper,the K and Na spectral information of different biomass fuel combustion flame is collected,and the spectral characteristics are extracted.Combined with the flame temperature,the prediction model of K and Na content and the tendency of fuel slagging is established.The main research contents are as follows.Firstly,the elemental analysis and industrial analysis of different biomass fuels are introduced.Based on the combustion experiments,the effects of fuel quantity and air quantity on the spectral intensity of K and Na elements in the combustion flame are analyzed respectively.Secondly,K and Na spectral information of different biomass fuel combustion flames are collected.After the data preprocessing and feature extraction,dynamic prediction model of K and Na content in biomass fuel has been build based on recurrent neural network(RNN),long short-term memory neural network(LSTM)and deep recurrent neural network(DRNN),which is compared with the static prediction model of the K and Na element based on SVM and BP neural network.The results show the effectiveness of the dynamic prediction model.Finally,the main factors affecting the slagging of biomass combustion are analyzed,including the melting temperature of biomass ash,the effects of K,Na and other elements,and the classification of different slagging degrees.The characteristics of spectral information of K,Na elements in biomass combustion flame are extracted.Combined with the flame temperature,and dynamic prediction model of biomass combustion slagging tendency based on RNN,LSTM,and DRNN are constructed,which are compared with the static prediction model of slagging tendency based on SVM and BP network.The results show that different prediction models have certain accuracy rate,and the accuracy of the online prediction model is higher than the traditional offline measurement method.