基于神经网络逆控制的钢包精炼炉的仿真研究

Simulation of the ladle furnace electrode system based on neural network inverse control strategy

基于逆动力学控制的思想,提出一种基于RBFN逆模型控制策略,并将该控制策略对具有多变量非线性特性的钢包精炼炉系统进行控制。采用三个RBF神经网络分别对被控对象的三个输入进行逆辨识,每相神经网络逆辨识模型反向作为逆控制器模型与每相串联,从而构成三个已解耦的独立的伪线性对象,进而针对三个独立的伪线性对象进行线性控制,从而实现了对三相耦合系统的精确控制。仿真实验证明了这种方法的可行性和有效性。

An inverse control strategy based on RBF neural network was presented in this paper.The LF electrode system with the characteristic of non-linear and multi-variable was controlled an decoupled by the control method.Three inputs of the plant were identified by three RBFNS.Three dynamic pseudo linear systems were formed through the models of controllers and the plants are in series to decouple into three independent pseudo-linear plant.So it can use linear method to control those plants.The simulation results show the feasibility and effectiveness of the strategy.