| The ball mill plays a major role in the pulverizing system of a thermal power plant. It is a complex controlled object which is multivariable, nonlinear, strong coupling, time-varying, pure delay and large inertia. Therefore, the conventional mill control system based on single-loop PID is difficult to put into use. In order to obtain satisfactory control effect, try to maintain ball mills in the optimum operating state on the premise of safe operation, namely ensure the larger output of ball mill while reducing consumption of power so as to improve energy efficiency of ball mills and enhance economy, security and reliability of unit operation. Study of this topic not only has important meaning in non-linear control of complex objects, but also meets the demands of the industrial field. And it has important practical significance for the development of energy efficiency.Based on mechanism model of ball mill, the problem of measuring the coal load of ball mill is solved by the method of fuzzy multi-model soft sensor based on neural network. At first, a mechanism model of ball mill is built. Then the implementation of fuzzy multi-model soft sensor based on neural network is deeply discussed, namely how to combine FCM clustering and multi-model theory based on RBF neural network. Then this method is applied to the simulation experiment of measuring the coal load of ball mill. Finally, on the basis of deep study of self-optimizing control, fuzzy neural network decoupling and fuzzy-PID control, propose overall strategy of mill’s optimal control.This thesis not only has proved the feasibility and validity of fuzzy multi-model soft sensor based on neural network in measuring the coal load of ball mill successfully, but also the proposed optimization strategy can maintain ball mills operating near the best point. It has certain reference value and consulting significance for solving optimal control problem of complex controlled object in practical production process, such as ball mills. |
PDF Full Text Request