论文部分内容阅读
管式加热炉具有典型的非线性、多变量、大时滞、强耦合和时变特性,传统的控制策略很难获得理想的控制性能。为此,提出了基于一类多模型的自适应神经网络预测控制方法,首先基于不同负荷下加热炉的运行情况建立多个自适应神经网络模型,预测变负荷、大扰动时的加热炉输入输出状况,然后通过自适应调整神经网络模型的结构和参数跟踪加热炉由于参数时变或其他干扰引起的系统漂移,最后应用粒子群算法对基于多模型自适应神经网络进行滚动优化,获得加热炉操作变量的次优控制律。此方法可以有效地跟踪多路进料、多燃烧器加热炉的控制指标,提高了加热炉的整体热效率,并且能够节约燃料,减少温室气体排放。所开发的控制系统成功应用于某炼厂常减压加热炉装置,取得了良好的效果。
Tubular furnace has the typical nonlinear, multivariable, large time delay, strong coupling and time-varying characteristics, the traditional control strategy is difficult to obtain the desired control performance. For this reason, an adaptive neural network predictive control method based on a multi-model is proposed. Firstly, a number of adaptive neural network models are established based on the operating conditions of the furnace under different loads to predict the input and output of heating furnace under variable loads and large disturbances Then adaptively adjust the structure and parameters of the neural network model to track the system drift caused by the time-varying or other disturbance of the heating furnace. Finally, particle swarm optimization is used to optimize the neural network based on the multi-model and get the operation of the furnace Suboptimal control law of variables. The method can effectively track the control indexes of the multi-feed and multi-burner heating furnaces, improves the overall thermal efficiency of the heating furnace, and can save fuel and reduce greenhouse gas emissions. The developed control system has been successfully applied to a refinery atmospheric and vacuum heating furnace device, and achieved good results.