Research On Spray Drying Model Of Ceramic Industry Based On Artificial Intelligence

The hot air makes short and full contact with the ceramic slurry to realize the rapid drying of ceramic particles in the spray drying tower.Spray drying is a gas-liquid-solid multiphase flow phase transition system in which temperature,humidity and velocity fields are coupled together.The heat and mass transfer process of the system is complex,with nonlinear and serious lag.In addition,the spray drying process is fast,it is difficult to observe the inside,and the quality of the produced particles(moisture content,particle size distribution,etc.)is difficult to guarantee.Ceramic companies usually need to monitor the spray drying equipment for 24 hours to prevent changes in various parameters from affecting the quality of the particles.Moreover,real-time monitoring equipment for particle quality is not widely used in ceramic enterprises.In view of the above problems,it is particularly important to establish a ceramic industry spray drying model that can predict the quality of particles under actual working conditions.On the one hand,it can provide guidance on various parameters of spray drying on the premise of ensuring quality.On the other hand,it can predict the quality index of particles in real time to meet production needs and avoid the generation of waste and reduce the waste of energy.This topic is based on artificial intelligence theory such as artificial neural network,intelligent algorithm,etc.,to analyzes the heat and mass transfer mechanism and influencing factors of particle quality in the spray drying process of ceramic industry.The ceramic industry spray drying quality prediction is established based on back propagation neural network.Introduce improved moth-flame optimization(IMFO)algorithm to improve model prediction accuracy and achieve real-time prediction of multiple quality indicators of ceramic particles.In the continuous production process,various parameters of spray drying can be adjusted in time to reduce energy waste.the main research contents are as follows:(1)Analysis of the drying process and heat budget of ceramic industry spray drying equipment.Moisture content and particle size distribution are particle quality indicators,with the inlet air temperature,nozzle aperture,initial moisture content,slurry specific gravity,inlet air velocity and feed velocity as the main factors affecting particle quality.Orthogonal experiment design is carried out on the basis of quality indicators and influencing factors,constructing a level table with 3 levels and 6 factors and using orthogonal table L27(313)to provide a data source for the construction of the spray drying artificial neural network model in the ceramic industry.(2)Analyzed the structure and mathematical principle of BP neural network in artificial neural network,combined with the characteristics of ceramic industry spray drying,established a BP neural network ceramic industry spray drying quality prediction model.It can be concluded that the overall average fitting degree between the predicted value and the actual value of the model is 0.7873,the average error is 2.7635,and the prediction accuracy is lacking.(3)Due to insufficient prediction accuracy of the model,the moth-flame optimization(MFO)algorithm was introduced,and multiple strategies were improved,such as good point set strategy,adaptive helix angle,and Levy flight strategy.Finally,an IMFO algorithm is formed.The MFO-BP neural network model and IMFO-BP neural network model of spray drying in ceramic industry were established respectively.The simulation results of the quality index prediction show that the prediction performance of the MFO-BP neural network model is an average increase of 6.13%to 0.8355,and the overall error is reduced by an average of 26.27%;the prediction performance of the IMFO-BP neural network model is an average increased of 12.07%to 0.8823,and the overall error is reduced by 32.61%.In addition,the absolute error rate of each model is divided into three ranges of 0-5%,5-10%,and>10%.The number of error rates predicted by the IMFO-BP neural network model falls within 0-5%is greater than the other two models,It proves that the IMFO-BP neural network model is more suitable for ceramic industry spray drying quality prediction,and has stronger adaptability.(4)Using MATLAB software based on IMFO-BP neural network ceramic industrial spray drying model to design a particle quality prediction platform,the main interface is composed of parameter input area,particle quality prediction area and prediction button area,which can provide users with real-time prediction and single prediction function help to guide the actual production.The subject provides a certain reference value for the ceramic industry spray drying quality prediction and the intelligent development.