Research Of Field Temperature Control Of Chain Grate

In the methods of pellet production, grate-rotary kiln, which have a better adaptability and the raw material cost is lower, is widely used in our country. Chain grate is an important equipment of grate-rotary kiln, and the temperature of it is the major factor that affects the quality of pellet. Now, the research of the temperature of chain grate includes the modeling of it, the controller of it and design of process parameters. And most research of the modeling of it is completed under condition that parameters are changeless. Then, the really situation of the temperature of chain grate cannot be reflected. On the other hand, the temperature of chain grate is a large dead-time controller device, and it is influenced by many parameters, so a conventional controller may not have a good control effect. In order to solve those problems, establish a model of chain grate’s temperature with the changed parameters, and design a better controller. The main contents are as follows:(1) Based on the first law of thermodynamics, mechanism of heat exchanged in pellet processing of chain grate is analyzed, and equilibrium equation is established. By finite difference method and MATLAB, model of chain grate’s temperature, which in a single time, is solve. And model of chain grate’s temperature, which in continuous time, is solve by mixed programming about Lab VIEW and MATLAB, then the simulation model is established. The simulation show that output this simulation meets the truth and has a higher accuracy.(2) Effect of quantization levels precision on learning precision and the shortcomings are analyzed based on description of mapping process of CMAC neural network. In order to solve shortcomings of CMAC neural network, the concept of membership of perceptions and weights are proposed. Then, mapping process of CMAC neural network and learning process of weights are improved. The simulations show that this new CMAC neural network can improve learning precision without increasing quantization levels precision.(3) Characteristics of temperature field of chain grate that the effects of hot gas’ first temperature, pellet’s first temperature and pressure on pellet’s final temperature are analyzed. Then, a feedforward controller is designed. The inputs of it are hot gas’ first temperature, pellet’s first temperature and pressure. The output of it is pressure. The specific algorithm of it is designed by combined with the new CMAC neural network and fuzzy. The simulations show that the pellet’s temperature can be controlled in setting range after off-line learning. And, the system can quickly respond to stability by on-line learning when parameters changed.