Research And Application Of Drying Control Model Based On BP Neural Network

Drying machine is the key equipment for processing workshop of cigarette factory.Due to previouse process,tobacco thread has a greater moisture content,and the moisture content of the tobacco will make be tobacco thread sticked with each other,so they are not fluffy enough to make volume.They must be baked by drying machine,so full of loose leaf silk,uniform moisture content,and have some curl,increase the filling capacity of tobacco thread,and reach the technological requirements.Therefore,it is a problem worthy of study to control drying machine's baking process,in order to improve the quality of tobacco products.At present,first dry and dry tail occurs at the processing workshop of tobacco drying process,and it is problem which is a common phenomenon in the tobacco industry right now.The main purpose of this research is to research and improvement Drying Machine Control System for the tobacco factory,control effects of moisture,reduce cut tobacco drying process dry head,tail dry volume,and reduce the consumption amount of thread for single box of cigarettes.Based on full investigation of existing domestic and international tobacco drying machine and its operating principle,control mode,we focused on the current structure of the present German companies HAUNI'KLK Drying Machine.The traditional control parameters and control methods were in-depth analyzed.Considering the limitations of conventional PID controller,multilayer neural network's forward and back propagation algorithm were introduced.The actual situation in the workshop production is combined with improved classic neural network algorithm to optimize parameters,including drying factor Ckl,sheet steam valve PI control parameters.So tobacco control baking process is optimized,and control accuracy of moisture after drying tobacco is improved.The main contents include:Introduction and analysis of relevant domestic and international status quo;the main neural network and its learning algorithms are introduced and analyzed;principles of conventional PID controller is analyzed,and the limitations of the analysis,and the combination of an improved method of neural networks.Finally,Matlab is used as simulation.Since the manufacturers' permission is needed for further improvement of the existing equipment,research result currently can not be directly applied to production process.In order to verify the validity of the algorithm,Matlab is used to vapor pressure-temperature conversion formula for the curve fitting and simulation.The empirical formula obtained within a certain pressure range,it improved the original steam pressure-temperature conversion algorithm and the tobacco control accuracy.It also significantly reduced too dry thread of feed head and end.Our method has a certain value in engineering.