Studies On Modeling And Optimum Control In The Coal-pulverized Storage System With Tube Ball Bill

Tube ball mill is the core of coal pulverized storage system, which is widely used in small or medium scale power plants home and abroad. It is a typical nonlinear multivariable coupling system. Conventional control strategy based on the linear object model, can hardly achieve ideal control effect. It's important to build a mathematical model of this system to realize the optimal control.In view of ball mill system control strategy, several multivariable system decoupling algorithms have been researched. The modeling method of complex system has also been described.Base on the operation mechanism of ball mill system, combining with previous achievements, new multivariable system control algorithm has been introduced. Specific modeling research has been done respectively with consideration of steel ball wear, mill load and the whole system. The simulation shows the effectiveness of modeling and optimal control.(1) The comprehensive analyses are carried out owing to the indispensable knowledge of the wear condition of the steel ball. The general mathematical model of the steel ball wear is set up,which not only describes impact of changes of the parameters of ball mill's working conditions, but also reflects the economy of the operation. Subsequently, the method of identification is presented for the model with pan-grey number linear equations and least square regression. This wear model has very profound theory value to analyze the economy operation on how to carry on the matches of the steel ball and when to add balls.(2) Based on Internal Model Control (IMC) principle, a new kind of PID arguments tuning method for multivariable system is proposed. The coefficients of proportional term, integral term and derivative term are directly obtained from the present method. Computer simulations in ball mill process of pulverizing system is done, and the results show that PID controller tuned by the new method is very superior to the general tuning method in the following aspects: control quality, robustness and disturbance rejection capability while controlled plant varies.(3) Making use of the CMAC neural network principle, the algorithm of soft sensor for CMAC is carried out. The sample data from the abnormal and normal conditions of the ball mill's are used to train the CMAC neural network. The fill level of tube ball mill is implemented by soft sensor. It is proved that the CMAC model can reflect the fill level preferably. It also provides the guide and convenience for control and optimization of pulverizing ball mill. It will reduce the electricity-using.(4) On the basis of the dynamic characteristics mechanism analysis of tube ball mill, using T-S fuzzy neural network, with the idea of simplified subtraction cluster and improved PSO-BP hybrid algorithm to optimize the the consequent parameters of rules, we model and identify the system of tube ball mill. The simulation results show that the algorithm based on improved PSO-BP hybrid model has the higher recognition accuracy than the algorithm which is the least-squares estimation based on the recurrence Kalman filter. The modeling method provides a reference for optimizing allocation and improving quality control of the ball mill control system, but also lays the foundation for the design of T-S fuzzy controller.Finally, a summary of the research results is addressed, and the future research of the modeling and optimum control algorithm used in the paper for practical project aspects of the power plant tube ball mill system is discussed.