摘要
针对钢包精炼炉(Ladle Refining Furnace)又称LF炉,配料加料过程的惯性、时滞、非线性等控制特性,设计了一种基于微粒群优化算法(Particle Swarm Optimization,PSO)、误差反向传播(Back Propagation,BP)神经网络以及比例-积分-微分(PID)的复合控制算法PSO-BP-PID,并将该复合算法应用于150 t钢包精炼炉配料称重控制系统中,实现配料称重过程的智能控制。PSO-BP-PID算法利用微粒群优化算法的全局寻优特性,优化BP神经网络的初始权值以提高神经网络的收敛性;采用经微粒群算法优化后的BP神经网络在线实时调整PID参数。通过基于PSO和BP网络的PID控制器实时控制钢包精炼沪的配料过程。仿真实验和运行实验结果表明,PSO-BP-PID算法的控制效果优于单一PID算法的控制效果。采用PSO-BPPID算法的钢包炉配料系统后,明显提高了配料精度,有效地解决了配料称重过程中速度与精度的矛盾。
In accordance with the control features of material proportioning process of the ladle refining furnace, e. g. , inertia, time lag, non-linearity, a kind of compound control algorithm is proposed based on particle swarm optimization algorithm (PSO) , error back propagation( BP)neural network and proportion integration differentiation( PID)algorithm. The PSO-BP-PID compound algorithm is ap- plied in a 150t ladle refining furnace burden weighing control system. The particle swarm optimization algorithm with global optimization characteristics improves the convergence of the BP neural network which the initial weights of BP neural network is optimized. The opti- mized BP neural network is then used to adjust PID parameters on-line. The PID controller based on PSO and the BP neural network controls real-time the proportioning process of the ladle refining furnace. The simulation and operation experimental results show that the control effect of the PSO-BP-PID algorithm is better than the control effect of the traditional PID algorithm. The control system of the la- dle furnace ingredients based on PSO-BP-PID algorithm can significantly improve the accuracy of ingredients, and effectively solve the contradiction between in^edients weighing speed and accuracy.
出处
《控制工程》
CSCD
北大核心
2013年第5期825-828,832,共5页
Control Engineering of China
基金
国家自然科学基金(11272119)
湖南省工业科技支撑计划项目(2011GK3160)
关键词
钢包精炼炉
配料称重
微粒群优化算法
神经网络
PID控制
ladle refining furnace
proportioning and weighing
particle swarm optimization
neural network
PID control