Study On Modeling By Neural Network And Intelligent Control Of Wood Drying

With the reduction of global forest resources day-by-day and the appearance of environmental protection and ecological problems, governments all over the world paid attention to these problems such as how to use the limited wood resources reasonably, reduce energy resource consumption and increase the quality of wood products. Faced by our country with fewer forest resources, how to improve using quantity and raise utilization ratio is one of the most important projects. Wood drying is an important measure for improving wood physical mechanics performance, utilizing wood reasonably, reducing loss of lower class and raising wood utilization ratio. It is one of the key technologies for guaranteeing the quality of wood products. Wood drying is a complicated non-linear process that is lagged, time-variable and coupling. For them, it is difficult to build up ideal drying model, which is corresponding to reality. It is difficult to control automatically because there are so many affecting factors. Traditional PID controlling method lacks sufficient flexibility. It is urgent to build up modern control system of high quality for innovative drying equipment.In the dissertation, the wood drying models were built up based on Neural Network. The Neural Network was identified by studying input and output data. It made the goal functions minimum, thus implying the relationship of input data and output data, namely the model describing system. The temperature-humidity controlling model and drying schedule model were built up with the Time Delay Neural Network (TDNN) and Dynamical Recurrent Neural Network (DRNN) which were suitable for identifying non-linear system. Temperature-humidity controlling model was the model that described the relationship among three controlling signals including heat-valve, spray-valve, eliminating-damp-valve and temperature and humidity, which offered the essential conditions for designing controller. Drying schedule model described the relationship between temperature-humidity and wood moisture content, which provided the mathematic model for drying schedule, predicted for the changing characters of moisture content and offered the effective basis for optimizing the drying standard. Drying schedule inverse model could provide the values of temperature and humidity in drying kiln at this moment according to the moisture content, thus making the stage drying in succession which vied moisture content schedule as drying schedule. In our study the structure of the Neural Network and learning algorithms were given, in addition, we trained and verified the model through experimental data. The simulation results showed that the models were effective and reliable. We compared the results of two kinds of Neural Networks and made the conclusion that identification effectiveness based on TDNN was better than DRNN and more suitable for describing the complicated object such as wood drying.The controlling course of wood drying was a process of controlling the temperature and humidity of the medium in kiln to make the wood moisture content reduce to a certain expected value. According to the characteristic of real dry course, we examined the utilizationof fuzzy control and fuzzy adaptive control in wood drying control system based on Neural Network model. The simulation result showed that the two kinds of control methods were better than traditional PID control. This was the basis for improving controlling level of wood drying and realizing controlling wood-drying course fully automatically. It was importance for guaranteeing the drying quality and reducing the energy consumption and the cost effectively.In order to prove the controlling results of the controller, we performed wood drying experiments with drying kiln. The results of operating actually showed that the control system could meet requirement of precision and had a good control results.