变风量空调室温滞后神经网络模型辨识及预测

Identification and Prediction of Room Temperature Delay Neural Network Model for VAV Air Conditioning

针对变风量空调系统室温大滞后动态响应特性数学描述问题,介绍了基于神经网络的模型辨识基本原理。基于Elman网络模型结构特点,提出了一种Elman网络层延迟系数最优选择算法,用以确定室温对系统调节量的滞后时间。在此基础上,进一步建立了基于Elman网络的室温时滞系统多步预测模型。通过仿真试验研究,验证了提出算法的准确性和有效性。

Aiming at the problem of mathematical description for dynamic response characteristic of indoor temperature time-delay system,the fundamental principle of neural network model identification is introduced in regulation process of variable air volume(VAV)air conditioning system.Considering the model structure of Elman neural network,this paper presents an optimal selection algorithm for layer delay coefficient in order to determine delay time between indoor temperature and regulation parameters;and a multiple-step prediction model of indoor temperature time-delay system based on Elman neural network is built.The effectiveness of the proposed method is validated through the simulation experiment.