Research And Application Of A Neuroevolutionary Horseshoe Flame Furnace Thermal Efficiency Prediction Model

Glass is an important basic material for the construction industry,automotive industry and other industries,and the glass industry is the country's livelihood industry.But at the same time,the glass industry consumes a lot of energy,energy saving and emission reduction is a huge challenge for the glass production enterprises.At present,the research on the optimization of the energy efficiency of the furnace mainly revolves around the optimization of the structure of the furnace,but for the enterprise,how to optimize the energy efficiency of the glass furnace equipment already in use is the most important issue of concern.In the working process of glass kiln,there are not only complex physical processes,but also complex chemical processes,so it is necessary to explore the system structure and production process of horseshoe flame kiln,conduct thermal balance analysis,establish thermal efficiency prediction model,and provide guidance for enterprises to make optimal production decisions.The research object of this thesis is the regenerative horseshoe flame glass kiln,and firstly,the influence of the kiln production process and kiln structure on the kiln combustion is investigated,and the thermal equilibrium model of horseshoe flame kiln is constructed.Data pre-processing for possible anomalies in the high temperature operating environment of the glass kiln reduces interference with the predictive model.The effect of data such as furnace pressure and temperature on the thermal efficiency of the furnace is analyzed by Pearson correlation coefficient to identify the main influencing factors and construct a neuroevolution-based thermal efficiency prediction model.Simulation experiments were then conducted in conjunction with actual production data.The final synthesis of the study applies the thermal balance analysis module and the thermal efficiency prediction module to practice.The main contents of this thesis are as follows.(1)This thesis provides an in-depth analysis of the system structure,process flow and combustion method of the horseshoe flame furnace in response to the complex process flow and high energy consumption in the production process of horseshoe flame furnaces.The analysis of the overall production frame to the local structure of the glass furnace provides theoretical support for the thermal equilibrium model of the regenerative horseshoe flame glass furnace.(2)For data abnormalities,missing,redundancy,etc.caused by harsh environment,network failure,sensor failure,etc.in the actual production environment,statistical process analysis and quartiles are used to identify data abnormalities at the same time,so as to meet the accurate identification of abnormalities while also meeting the timeliness of data processing.Combined with the theory of thermal equilibrium,the energy input and energy output of the furnace were calculated,and a thermal equilibrium model was constructed,resulting in thermal efficiency values,which provided a data base for subsequent thermal efficiency predictions.(3)Building thermal efficiency prediction models.This thesis presents a thermal efficiency prediction model based on neural evolution.In order to construct evolutionary neural networks more easily and quickly,evolutionary strategies were used as neural evolution methods.Combined with the chronological nature of the glass furnace data,the evolved neural networks use long-and short-term memory neural networks.In order to speed up neural evolution,a novelty reward approach is proposed for the evolution of neural networks.To examine the models' predictive accuracy,this thesis uses BP neural networks and long-and short-term memory neural networks for comparison experiments.Experimental results analysis,thermal efficiency prediction models based on evolutionary nerves perform better.Combining the above research methods and models,on the basis of the original energy management system,using the Java language to expand and upgrade development.The thermal equilibrium model and thermal efficiency prediction model are embedded in the system and the system is eventually put into operation in actual production.