【作者】 夏志峰；

【导师】 韩宁； 王亚慧；

【作者基本信息】 北京林业大学 ， 控制理论与控制工程， 2012， 硕士

【摘要】 随着建筑物结构和功能复杂程度的增加以及建筑智能化程度的提高,室内空气品质的舒适度越来越依赖于空调系统的控制。而在空调系统控制中,室内空气品质参数的有效性和室内空气品质模型的精确性则是影响系统控制效果的关键因素。目前,室内空气品质参数检测存在着布线麻烦、检测范围小等缺陷,而且室内品质模型的精度往往不能达到控制系统的要求。为此,作者参与了北京市自然科学基金项目(8072008,Z07004)开发了基于无线传感器网络的室内空气品质检测系统,研究了基于多变量系统辨识理论的室内空气品质模型。具体内容如下：1、无线传感器网络方面1)基于JN5148无线通信模块设计制作了无线传感器网络节点硬件电路,包括无线通信模块的外围电路、电源电路和功能模块电路。2)基于PIC单片机设计了硬件定时控制电路,采用控制负载电源通断的方式降低了近70%的节点平均功耗。3)基于Zigbee PRO协议和JenOS操作系统开发了的节点软件应用程序,实现了协调节点、路由节点和底层节点的相应功能。4)基于ARM9开发板和WinCE操作系统开发了网络的主控制器用于提高协调节点的处理能力。2、室内空气品质建模方面1)提出了基于微粒群-滑模变结构控制(Particle Swarm Optimization-Slide mode Control, PSO-SMC)的神经网络训练算法,相比于单一的滑模变结构控制算法,降低了网络的训练误差,提高了网络的泛化能力。2)针对模型的结构辨识,采用Guidorzi方法和神经网络时滞算法分别估计了模型的阶数和时滞系数。3)针对模型的参数辨识,通过PSO-SMC算法训练的神经网络估计系统模型参数,并最终得到了拟合度较高的室内空间空气品质模型。更多还原

【Abstract】 As the complexity of the building structure and function increases, as well as building intelligent improves, indoor air quality is increasingly dependent on the control of the air-conditioning systems. In air-conditioning control systems, the effectiveness of indoor air parameters and the accuracy of indoor air quality model are key factors which effect the control result. At present, there are many problems in the detection of indoor air parameters, such as route complexity, small test range. And most indoor air quality models can not meet the qualification for control system. Therefore, the author of this paper developed a wireless sensor network based indoor air parameters detection system and identified a multivariable system identification based indoor air quality model. The research was supported by Beijing Municipal Natural Science Foundation. The main taks are listed as follows:1、In the wireless sensor network aspect1) Designed the JN5148based circuit of wireless sensor network node which include communication module’s peripheral circuit, power supply circuit and function module circuit.2) Designed the PIC based hardware timing control circuit which controls the make-break of load switch in a assigned time span.3) Developed the Zigbee PRO and JenOS based node software programs which realize the corresponding functions of coordinator, router and end device.4) Developed the ARM9and WinCE OS based main controller which improves the coordinator’s processing ability.2、In the indoor air quality identification aspect 1) Proposed a PSO-SMC (Particle Swarm Optimization-Slide Mode Variable Structure Control) based artificial neural network training algorithm. Comparing with SMC, this algorithm reduces training error and improves network’s generalization ability2) In order to determine indoor air quality model’structure, Guidorzi and neural network time delay identification algorithm were used to identify system’s order and time delay parameter.3) In order to determine indoor air quality model’coefficient, PSO-SMC algorithm was adopted to estimate model’paremeters. Finally, indoor air quality model is established with a high fitting degree.更多还原