基于神经网络模型的锅炉广义预测控制

A Neural Net Model-based General Predictive Control Strategy for Use on Boilers

针对目前电站锅炉运行中 ,控制性能不佳 ,热效率低下等不足 ,提出了运用基于神经网络模型的GPC控制方法(NNGPC)以改善控制器性能的策略 ,并通过大量的仿真研究对其进行了验证。另外 ,为简化模型结构 ,便于实时在线计算 ,又利用改进后的Elman网络模型代替原有的多层前向BP网络模型进行了对比实验 ,效果令人满意。最后 ,利用仿真结果 。

In an effort to rectify deficiencies commonly encountered during the operation of current utility boilers, such as poor control performance, low thermal efficiency, etc., the authors have come up with a neural net model based general predictive control strategy to improve the relevant control device performance. Through numerous computer simulations the proposed control strategy has been adequately verified. Moreover, an improved Elman network model was utilized to replace the original multi layer feedforward model in order to simplify model configuration and facilitate on line real time calculations. A contrast test of the above two models shows that a satisfactory result in terms of effectiveness has been attained through the use of the improved Elman network model. Finally, on the basis of the simulation results, expounded were the selection of the parameters of the neural net model based general predictive control and some specific issues in engineering applications.