摘要
针对热网非线性、时变、大滞后特性,结合神经网络和一般广义预测控制算法,通过改进BP神经网络解耦器将来自其它通道的耦合影响视为可测干扰进行补偿,解决供热系统中存在的耦合问题;设计了神经网络解耦控制器来调节一次网的流量分配,进而间接调控二次网各用户的室内温度,提高热网的供热质量,达到稳定和均衡供热.Matlab仿真结果表明,在二次网用户初始温度相差越小的情况下,神经网络算法所需的调节时间越短,达到的控制效果越理想,即使在二次网各用户初始温度相差很大,经过一段较长时间的调节,也能够逐渐实现均匀供热.
The system parameters of heating network have a feature of nonlinearity,time-variant and high time-delay.Besides,the process parameter of hot media varies as load changes because of the coupling of adjustment process of heat exchanger station subsystem.Therefore,it is difficult to establish a mathematical model,for the heating network and traditional control theory can hardly be applied.According to the features of the heating system,an improved BP algorithm is proposed in this paper based on the comparative analysis of BP neural network algorithm and general predictive control algorithm.The improved algorithm regards the coupling effects from other channels as measurable interference and compensate for it,which minimizes the coupling effects.A comparison of the basic algorithm and the improved one is provided by simulation using Matlab.Its result shows the effectiveness of the improved algorithm.
出处
《哈尔滨理工大学学报》
CAS
2012年第2期58-62,共5页
Journal of Harbin University of Science and Technology
关键词
交互耦合性
神经网络解耦器
智能复合算法
interaction coupling
neural network decoupler
smart composite algorithm