基于系统辨识的太阳能—相变蓄热新风供暖系统控制策略研究

【作者】 李鹏；

【导师】 周晓光； 杨军；

【作者基本信息】 北京邮电大学 ， 机械电子工程， 2011， 博士

【摘要】 随着不可再生资源的日益减少,低碳节能、用绿色能源代替传统能源已经成为全世界要共同面对的问题；而随着科技的进步与人类文明的发展,人们对自身生活环境的要求也越来越高。如何协调处理能量损耗的降低与生存质量的提高之间的关系成了当今各国专家学者研究的重点,而太阳能-相变蓄热新风供暖系统就是在室温控制领域解决这两者矛盾的一种有效的新方式。太阳能-相变蓄热新风供暖系统综合考虑建筑节能、新能源利用、能源储备、室内通风需求等问题,提出利用太阳能技术、相变蓄热技术与现代的控制理论相结合的方式,解决室内的供暖和通风换气问题。系统能够实时监测现场各种条件的变化；根据现场的不同实际情况,把太阳能用于室内供暖与能量储备；根据用户需求实现加热室温、储存热量、释放热量之间的自动转换；并在每种转换模式中实现自动控制,使室内的空气质量、温度条件,达到最宜居的状态,创造最理想的生活环境。纵观国内外本领域的相关研究成果,从太阳能-相变蓄热新风供暖系统的系统特性出发、对它的能量转化、运行机制、系统模型、智能控制策略等一系列问题进行成体系的研究,是一个目前研究基础相对薄弱,但很有研究价值的新领域。本文的主要研究内容和成果包括：1.从热力学的角度分析了系统的导热流程和工作原理；并从功能结构的角度对系统进行分析,把系统分为导热系统与测控系统两大部分。随后分析了导热系统的整体设计；测控系统中硬件平台主要测控参数的确定、测控设备的选用、测控网络结构的设计；以及软件平台中数据采集、数据处理、数据监视和控制输出等几大功能的设计。2.对太阳能集热器、蓄热槽、风机盘管进行物理特性研究,并根据其物理特性进行数学模型的推导；得出了系统各部分热传递的特性,同时对控制策略的模拟对象进行了研究。其中,用来加热新风的风机盘管是控制系统的主要控制对象,所以在它的数学模型基础上进行了仿真研究,并用实验数据对数学模型进行了验证。3.分析太阳能集热部分的控制需求,针对不同的室外光照条件提出了两种控制模式。对蓄放热、新风加热部分的控制需求进行分析,并针对不同的导热量提出了既能满足新风供暖需求,又能最大限度节能的五种控制模式。随后分析讨论了模式转换的条件,并设计了系统整体的控制模式转化策略。4.对非线性系统的辨识方法进行了分析；经过辨识方法的比较,选择用人工神经网络中的BP网络作为本系统的辨识方法。通过对BP网络的网络结构、数学算法及训练过程等问题的研究,确定了本系统的系统辨识策略。然后在系统辨识阶段比较并确定了神经网络辨识结构、BP网络的各层结构、网络训练算法和改进算法,得出了系统各个控制模式下的BP网络辨识模型。最后通过仿真验证了模型的有效性。5.对几种基于模型的控制理论进行了分析和对比,选择用模型预测控制作为本系统的控制算法。随后分析了基于线性模型的预测控制理论与基于非线性模型的预测控制理论,并研究了模型预测控制的数学描述以及算法特点。继而针对本系统的特点,进行神经网络预测模型与模型预测控制器的设计。最后对系统的各种控制模式进行了恒温测试与随动测试的仿真,并对仿真结果进行了分析,验证了控制策略的可行性。更多还原

【Abstract】 With the reduction of non-renewable energy, substitution of low-carbon and green energy for traditional energy has become the worldwide concern. Moreover, with the progress of science and development of civilization, higher requirements are put forward for the comfortable living environment. How to balance the relationship between low energy consumption and high living standard becomes the focus of researchers. Nowadays, fresh air heating system with scholar energy and phase change thermal storage provides an effective solution for the above contradiction in indoor air temperature control field.In such a fresh air heating system, issues of building energy conservation, new energy utilization, energy storage, and ventilation requirements are all considered. In addition, combination of techniques such as solar energy, LHTS, and modern control theory is proposed to solve the indoor heating and ventilation problems. In the fresh air heating system, the following functions are realized:real-time monitoring of various condition changes; automatic shifts among air heating, energy storage and energy release according to occupants requirements; automatic shifts among different modes to achieve an ideal living environment with high air quality and comfortable temperature. From the review of relevant domestic and international literatures, studies based on solar energy and LHTS fresh air heating system characteristics, such as energy change, operation mechanism, system modeling, and intelligent control, are rare, which are of great research value but with weak study basis.The main achievements of this dissertation are as follows:1. From the view point of thermodynamics, thermal conductivity process and work principle of the system are analyzed. The system is divided into two subsystems from the view point of system function:thermal conductivity system and test and control system. Then the following contents are introduced:design of thermal conductivity system; determination of parameters for the test and control system hardware platform, selection of the test and control devices, design of the test and control network; realization of the data collection, data processing, data monitoring and output control functions.2. Physical characteristics of solar collector, heat storage tank, fan coil units are studied. Through mathematical model analysis of the physical characteristics, thermal conductivity laws of different parts of the system are obtained. Meanwhile, the simulation objectives for control strategies are also studied, among which, fan coil unit is the main controlled objective. Thus, the mathematical model of fan coil unit is simulated for study and experimental data are used to validate the model.3. Control requirements of the solar collector part are analyzed. Then two control modes are proposed according to different outdoor light conditions. Then control requirements of heat storage and release and fresh air heating parts are analyzed. Five control modes are introduced for different heat conducting amount, which meet the requirements of fresh air heating and energy conservation. Subsequently, conditions for mode shifts are discussed, and control mode shift strategies for the whole system are designed.4. Methods of nonlinear system identification are briefly introduced. Through comparison of identification methods, BP neural network is selected as the system identification method. After analysis of BP neural network structure, mathematical algorithm and training process, BP network identification process is determined. During the system identification, neural network structure is chosen and the different layers of BP network are set. After determining the network training algorithm and improved algorithm, BP network identification models under different control modes are obtained. Then simulation and validation of the models are carried out.5. Control theories based on models are analyzed and compared. Then the MPC (model predictive control) is introduced whose control properties are better. MPC theories based on both linear and nonlinear models are analyzed, and MPC mathematical description and algorithm characteristics are studied. In view of the system features, a neural network MPC controller is developed. Invariable temperature tests and follow-up tests are simulated by various control modes. Simulation results are analyzed and show that the control strategies are feasible.更多还原