摘要
针对热力站为多变量、非线性、强耦合、大时滞的复杂时序控制系统,难以建立精确模型的问题,提出基于循环神经网络的长短时记忆算法对热力站控制系统建模,该算法既考虑到时间上的影响因素,又解决了长序列信息丢失的问题。以包头某热力站大量实时工况数据通过tensorflow框架搭建神经网络模型,仿真对比结果表明,长短时记忆网络建模能有效地减小建模误差,进一步提高神经网络在热力站系统建模中的精度。
Aiming at the problem about the heating station control system belongs to multivariable, nonlinear, strong coupling and large time delay complex process control system, it is difficult to establish an accurate model. The long short-term memory based on recurrent neural network is proposed to model the heating station control system. The algorithm not only considers the influence factors in time factor, but also solves the problem of long sequence information loss. Using a large real-time data in Baotou heating station, the neural network model is build by the tensorflow framework. The simulation result shows that the long short-term memory network modeling can effectively reduce the modeling error, and improve the accuracy of neural network modeling system in heating station.
作者
李琦
于明伟
LI Qi;YU Mingwei(College of Information Engineering,University of Science and Technology of Inner Mongolia,Baotou,Inner Mongolia 014000,China)
出处
《计算机工程与应用》
CSCD
北大核心
2018年第24期227-233,共7页
Computer Engineering and Applications
基金
国家自然科学基金(No.61463040)
