Study On The Theory Of The Predictive Control Of Maize Drying System Based On Fuzzy Neural Network

The grain drying is a nonlinear, time-delay and complicated exchange process of heat and substance, which is influenced not only by the characteristics of grain but also by climate. It is also a biological and chemical process besides physical one. With the guidance of the characteristics of grain, the problem is unsolved that how to predict the outlet moisture and control the discharge rate. On the view of thermodynamics, the grain drying process is an open thermodynamics system. The exchange degree of heat and substance decides the interior state of grain particle immediately. The temperature variety that embodies the interior state of grain can be observed, and it is also an important parameter of heat and substance exchange during the drying process. In total, according to the laws of temperature variety during the drying process, the methods of predicting the characteristics of maize drying process and using the laws of temperature variety to adjust the maize discharge rate are the primary issue to solve the problems in the maize drying process. They can overcome the deficiency which caused by the system time-delay and supply efficient dependence for controlling the outlet moisture of grain online.The mathematical model of maize drying is decribed, and the the temperature distribution and the physical mechanism of moisture moving within maize kernels are analyzed. The experimental setup is maize dryer with four stages. It is studied separately that the temperature variety during the maize drying process. Then the moisture predictive recurrent fuzzy neural network (MPRFNN in short.) is constructed for predicting the outlet moisture of the given samples. Finally, according to the characteristics of maize drying, the discharge fuzzy controller is designed. The works in this paper are decribed as follows:(1) All the experimental data are acquisited at Qingyuan Grain Store of Liaoning Province in 2005 and 2006. Then arithmetic of calculating the temperature and discharge rates of the given samples is given out in this paper.(2) There are four assumed models of the micro description of physical mechanism of grain drying. According to them, the model of maize kernel temperature variety during drying process is proposed formally. The model can be described as:the exterior temperature is the drying air temperature, but the kernel temperature rises from the interior to the exterior; the water is deported and evaporated in the kernel interior, then moves towards to the kernel surface along the capillary channels and finally overflows the kernel. The experiments proved that:â‘ During the drying prophase, if the penetrability of grain scarfskin is very high, it is hard for the water in the interior to transgress out. When the samples enter the tempering stage, the vapor will condensate into liquid water and give out energy at meanwhile. As a result, the temperature of maize kernels at the tempering stage is higher than that at drying stage. On the contrast, if the penetrability of the grain scarfskin is low, the result is opposite accordingly.â‘¡During the drying anaphase, the resistance of the capillary channels in the maize kernels is the primary factor which influences the maize temperature. The temperature of maize kernels at the temper stage is higher than at drying stage according to the same principle as above.â‘¢If the maize kernels are frozen, the penetrability of the grain scarfskin is destroyed. According to the same principle as above, the samples that the temperature of the dried grain at temper stage is lower than that at corresponding drying stage.(3) The outlet moisture predictive fuzzy neural network is constructed. Then the samples data is emulated and the results are consistant with the experimental results well. It is found that the predicted results are almost the same as the actual results when the quantity of the training samples achieves a certain number.(4) Based on the above achievements, a new scheme of maize drying control is proposed, in which the characteristic of maize drying is predicted with the data of the former three drying stage and the drying time of the fourth drying stage is controlled by the discharge rate fuzzy controller. This method solves the problem of system control with time-delay.