基于神经网络的实验室排风系统静压开环控制方法
Neural network-based open-loop control method for static pressure of laboratory exhaust systems
摘要：
科研实验室变风量排风系统运行调节范围大,针对传统变风量系统中静压控制方法产生的系统振荡和能耗较高的问题,本文提出了一种新型实验室排风系统控制方法。采用人工神经网络的方法预测最优风机运行频率,既降低能耗又避免闭环控制的系统振荡。以某科研实验楼3层一实验室为研究对象,基于CFD模拟与实验测量2种方法分别获取部分工况下的最优风机静压值构成训练数据集。利用反向传播(BP)神经网络模型学习得到控制模型。经CFD与实验验证,预测值与模拟最优值的误差控制在2.08%以内。相较于传统定静压控制算法,本方法显著降低了系统运行压力,最大降幅可达142.8 Pa,节能率最高可达23%。
Abstract：
The variable air volume (VAV) exhaust system in scientific research laboratories has a large operating adjustment range.To address the issues of system oscillation and high energy consumption caused by traditional static pressure control methods in VAV systems,this paper proposes a new control approach for laboratory exhaust systems.The method employs an artificial neural network (ANN) to predict optimal fan operating frequencies,thereby reducing energy consumption while avoiding oscillations inherent in closed-loop control.Taking a laboratory on the third floor of a scientific research experimental building as a case study,CFD simulations and experimental tests are conducted to obtain optimal fan static pressure values under partial operating conditions,forming the training dataset.A back propagation (BP) neural network model is utilized to develop the control model.Verified by CFD and physical experiments,the error between the predicted value and the simulated optimal value remains within 2.08%.Compared with conventional constant static pressure control algorithms,this method significantly reduces system operating pressure (by a maximum of 142.8 Pa) and achieves up to 23% energy saving.
Keywords:laboratory; exhaust system; neural network; variable air volume (VAV); open-loop control; static pressure