【作者】 王铁权；

【导师】 井元伟；

【作者基本信息】 东北大学 ， 控制理论与控制工程， 2009， 硕士

【摘要】 传统空调器通过温度传感器感受室内温度变化来控制压缩机的启动和停止来实现对温度的控制。它对温度的调节是一种断续的变化过程,它的缺点是不能根据环境温度变化及时调整空调器的工作状态。人们为了解决传统空调在室温波动时压缩机工作状态频繁切换的问题,将模糊控制算法引入到空调系统的控制中来,使得系统的自调整性有了很大提高。该系统能够根据室内温度变化,通过对压缩机进行连续、动态、实时地变频调速来调节室内温度。尽管常规模糊变频空调与传统空调相比,技术上有了很大进步,但是由于隶属函数和控制规则的获取经常依赖专家经验,而专家经验的正确与否经直接影响空调的控制效果。另外,当隶属函数和控制规则确定之后,一旦环境温度发生突变,空调器工作状态仍然按照既定的隶属函数和控制规则进行变化,中间没有过渡区,造成的温度突变会使人感到不适。根据神经网络自学习能力强的特点,本文尝试采用将模糊控制与神经网络相结合的方法对压缩机进行控制,解决模常规模糊控制存在的不足。论文内容如下:首先分析了传统空调控制方法和常规模糊变频空调控制的优点与不足之处,对国内外的研究现状及发展趋势进行了综述,提出了智能变频空调模糊神经网络控制系统设计与实现的研究思路;通过对已有智能变频空调设计方案的比较,结合国内实际情况,给出了一种新的控制方案,并对其硬件控制系统及算法进行了初步设计。在此基础上,采用模糊控制和神经网络控制方法相结合的实现手段,给出了智能变频空调系统的模糊神经网络控制算法。根据变频空调控制过程的非线性与时变性特点,以温度偏差和温度偏差变化率作为输入变量、变频压缩机工作频率变化作为输出控制量,设计了变频空调模糊神经网络控制器和预测器。通过对隶属函数和控制规则的调整,提高了变频空调系统的智能化程度。仿真结果表明了所提方法的有效性,实现了变频空调的智能控制。根据智能变频空调对硬件控制的要求,分别设计了室内机和室外机控制系统。依据硬件控制电路设计了相应的软件。通过实验研究,优化了相关的控制参数,同时也验证了本文方法的可行性。最后对全文内容进行了总结,并提出本设计存在的不足及需要研究改进之处。更多还原

【Abstract】 Temperature control system of older air-conditioner depends on its sensors to monitor indoor temperature’s changes and to control the compressor to run at a certain speed. This kind of air-conditioning system is a discontinuous process to adjust temperature and its shortcoming is that it can not adjust compressor’s speed according to the changes of indoor temperature inevitably. In order to overcome PID’s shortcoming ,people have put a new fuzzy control algorithms into practice so that the system’s regulative quality has been improved greatly. The control system, named fuzzy variable frequency air-conditioning system, has the intelligence character to adjust temperature automatically and continuously according to temperature’s changes indoors. All this will avoid frequent switches of the compressor which always appears on the traditional air-conditioning system.Although fuzzy variable frequency air-conditioning system has made great improvement in technic compared with the traditional control method, it has a main problem, that is, its control rules are set by expert’s experience beforehand, mostly under standard environment. When indoor environment changes suddenly, the fuzzy variable frequency air-conditioning system is controlled sharply at set membership functions and control rules. It will have a great effect on comfortability of human.Because the neural network has the high ability of automatically learning,the auther tries to combine the fuzzy control and neural networks to control the compressor. The contents of the thesis are as following:The project designs the fuzzy neural networks controller (FNNC) and neural network predicter (NNP), which employes the error of temperature and the changing rate of temperature as input, the working frequency of compressor as output. The control system improves the variable frequency air-conditioning system’s intelligence by adjusting subordinate function and control rules. The air-conditioner’s working state can adjust automatically according to the dweller’s request and the changes of temperature.Firstly, the auther introduces the advantage and shortcoming of traditional air-conditioning control system and fuzzy variable frequency air-conditioning system. Then the auther introduces the current research situation both home and abroad, and raises the new project "Design and Realization of Intelligent Variabale Frequency Air-Conditioner Control System with Fuzzy Neural Networks". Based on it, the schema of the control system is determined and design the hardware control system briefly.After studying the method of fuzzy control and neural network, the auther summarizes the combination method of of the two theories and designs the fuzzy neural networks controller for the air-conditioner. The fuzzy neural network controller and neural network predicter are designed according to the character of the nonlinear and time-varying of the control procedure and it is simulated. The result proves its intelligence. The control system and software of the indoor machine and the outdoor machine are designed respectively to meeting the demand of the air-conditioning system to the hardware. The experiment studying makes a research on the experiment and optimizes some parameters, which proves the availability of the theory. At last, the auther summarizes the contents of the dissertation and bring forth the questions that needed to solve.更多还原