由于变风量空调系统存在着难设计、难控制和难调试的问题，使得它难以得到广泛使用。而这些问题的根源在于变风量空调缺乏良好的控制系统和控制方法。所以变风量空调系统的控制成为目前空调界和控制界的研究热点。

本文以我校智能控制变风量空调实验室为实验基地，实验室的控制系统采用美国Andover Controls公司的全套DCS系统，从现场仪表到组态软件都是比较先进的设备；空调系统选用开利公司的39F230整体式空调机组和泛亚公司的PA022W水冷式冷热水机组。这些配置代表了目前变风量空调技术的先进水平。本文在此基础上，以变风量空调系统控制方法的比较和分析为研究重点。

本文首先介绍了变风量空调系统的定义、组成和控制方案；结合实际条件，设计了本实验室变风量空调系统的控制方案。

其次，根据相关文献，建立了变风量空调房间模型，模型为一阶有延迟的对象，为控制算法的仿真研究提供了被控对象；根据PID控制、模糊控制、神经网络技术的原理，设计了PID算法和模糊神经网络算法。

最后，在MATLAB的平台上，通过Simulink设计和M函数编程，仿真了空调房间温度PID算法和模糊神经网络算法的控制效果。并在实验室基础上，运用Continuum软件组态、编程，对变风量空调房间温度实现这两种算法的实际控制。通过对实验结果的分析和比较，发现相比PID算法，模糊神经网络算法的控制效果更为优越。

Because VAV is difficult to design, construct and debug, it is difficult for VAV system to be used widely. The real reason for this problem is lack of good control system and control method. So, the VAV’s control is the ongoing hotspot of the air conditioning and automation field.

My research is based on IACL (intelligence air conditioning lab) of USST (University of Shanghai for Science and Technology). The lab’s control system is DCS (distributed control system) aided by Andover Controls Co.Ltd. These series of equipment, from local instrument to configuration software, are very advanced. The lab’s air conditioning system is AHU (air handle unit) type of 39F230 donated by Carrier Co. Ltd and chilled-heated system type of PA022W is aided by Fanya Co.Ltd, which are the most advanced technology. The focus of my research is to compare and analyses VAV’s control method.

At first, I introduced the VAV’s definition, VAV’s sub-system and its control system, and I also design the control strategy of each subsystem.

Then, according to related information, I create air-conditioned room temperature’s mathematics model depending on our air-conditioned room’s condition, which is a transfer function with time-delay, long-inertia. This model is the base for the further simulation research. After studying classic control, fuzzy control and ANN (artificial neural network) knowledge, I designed PID arithmetic and ANFIS (adaptive neural fuzzy inference system) depending on simulation and real need.

Thereafter, I simulated these two control arithmetic in the MATLAB platform, and find the effect of ANFIS is better than that of PID.

At last, using our configuration software ——Continuum to design interface and program, I tested these two control arithmetic in real VAV system control. After researching the control effect, I find that the effect of ANFIS is better than that of PID, which is according with simulation conclusion.