【作者】 白燕；

【导师】 任庆昌；

【作者基本信息】 西安建筑科技大学 ， 智能建筑环境技术， 2013， 博士

【摘要】 变风量中央空调系统作为智能建筑的重要组成部分，凭借其舒适、节能、灵活性等方面的优势被广泛关注。而现场工程及控制过程中系统的运行由于受到建筑物环境及空调负荷变化等因素的影响，使得各设备的运行偏离了所设计的最优工作点，从而影响系统整体的运行和节能。如何从系统全局出发，根据现场运行情况，构建系统稳态模型，搜索全局最优设定点，在满足室内空气品质和舒适性的前提下，最大程度地降低系统能耗，成为一个有意义的研究课题。对于中央空调这个由多个子系统组成的大系统，在局部控制优化的基础上，实现全局系统优化控制，将局部控制环节整合起来，从大系统的角度综合决策、协调优化是一种行之有效的手段。论文针对节能环保、室内空气品质及人体舒适度的综合需求，寻求系统全局层面的解决方案。论文基于大系统“分解-协调”理论与优化控制策略，在对智能建筑变风量(VAV)中央空调系统进行递阶结构分解的基础上，介绍实验系统的整体软硬件平台及各子系统的功能设计。分别从控制算法和节能策略两个角度进行研究。建立各子系统动态模型，并设计广义预测控制器及神经网络PID控制器用于底层子系统的控制。实现了基于广义预测控制算法的风系统变静压控制策略及基于NN-PID控制算法的需求控制通风策略。实验表明，算法具有较强跟踪性及抗干扰能力的同时体现出可观的节能潜力。引入室外气象参数用于空调负荷的预测，提出一种修正的ASHRAE系数法预测室外逐时温度，由此构建训练数据集。设计Elman神经网络及Grey-NN神经网络预测算法，对空调系统动态负荷的预测结果为全局优化目标函数及约束的确定提供依据。研究变风量空调大系统的稳态优化问题，建立系统稳态模型；构建全局系统优化运行工况模型；设计改进的关联平衡法(IBM)对全局系统进行协调优化，可在最优点处使关联达到平衡，且保证协调的收敛性；在设定的优化周期内，通过两级之间的交互，寻得各子系统的优化设定值，进而送至底层控制器，完成递阶优化控制。以冬季工况为例进行的实验结果表明，采用全局优化策略能够很好解决中央空调全局系统的控制和优化问题，具有一定的节能潜力，且可推广到一类大范围工况过程系统。更多还原

【Abstract】 As an important part of the intelligent building, VAV central air conditioningsystem is of widespread concern due to its advantages of comfort, energy saving andflexibility. But in the engineering and controlling process, the actual operation of eachdevice may deviate from the optimal design, thus affecting the operation and energyefficiency of the overall system. Therefore, it’s of high significance to constructsteady-state model, search global optimal set point according to the actual situation, andminimize the energy consumption on the basis of indoor air quality and human comfort.For central air conditioning system composed of multiple subsystems, it is aneffective means to make integrated decision from the large-scale systems based on localcontrol optimization. The thesis focuses on the comprehensive needs of energy saving,indoor air quality and human comfort, to explore the overall solution to the system.In the research, the VAV air conditioning system is decomposed in hierarchicalstructure on the basis of large-scale system decomposition-coordination theory andoptimal control strategy, and the overall hardware and software platform of theexperimental system and the design of its subsystem function are introduced.The research was conducted from the perspectives of energy-saving strategies andcontrol algorithms. The subsystem dynamic model was constructed and the GPCalgorithm and Neural Network-PID control algorithm for local control unit wasdesigned. Herein, variable static pressure control strategy and demand controlventilation strategy were implemented based on the generalized predictive controlalgorithm and NN-PID control algorithm respectively. The experiments show that the algorithm with strong tracking and anti-jamming capability exhibits the considerablepotential for energy saving.With the outdoor meteorological parameters adopted to predict the air-conditioningload, ASHRAE correction coefficient method was proposed to predict the outdoorhourly temperature. Thereby the training data set was constructed. The Elman neuralnetwork and Grey-NN neural network prediction algorithm were designed, and theforecast results of air conditioning system dynamic load could be the basis for theobjective function and constraints of the global optimization.Steady-state optimization problem of VAV air conditioning system was studied,and the steady state model was established in the research. The global systemoptimization operating conditions model was constructed. An improved interbalancemethod (IBM) was designed to coordinate the global system, and the association couldbe equilibrium in the optimal point to ensure the convergence of coordination.Energy-saving optimization of the system control was implemented according to thesearched optimized value in each optimization cycle through the interaction between thetwo levels.The experimental results show that, using global optimization strategy can be agood solution to the central air-conditioning control and optimization of the globalsystem with large energy saving potential in winter conditions. And the strategy can beextended to a class of large-scale conditions process systems.更多还原