一种基于混合建模的模块化神经网络控制方法

A Modular Neural Networks Control Method Based on Hybrid Modeling

针对传统神经网络用于复杂过程系统的控制时难于收敛的问题,文章提出了基于混合建模的模块化的神经网络模型。采取运行机理建模和神经网络建模相结合的方式,把输入样本空间进行划分,实现基于混合专家网络的建模。试验结果表明,对大型燃煤锅炉供热系统,文章提出的方法可以较好地提高供热系统的稳定性和供热质量。

Against difficult convergence problem when the traditional Neural Networks are applied to complex process system,a hybrid expert neural networks model is brought forward.In the new method,input samples are divided into sub-space and hybrid expert neural networks model is realized.The experiment indicates that better stability and quality can be obtained through the new neural networks model in the huge coal-burning boiler heating system.