以人为本的智能家居辅助决策系统的研究与实现

发布时间：2018-05-02 14:07

本文选题：物联网 + 智能家居　；参考：《吉林大学》2015年硕士论文

【摘要】：近几年来，随着计算机网络技术的不断发展以及与其他技术的交叉应用，物联网(Internet of Things)的概念被提出并受到高度重视。现阶段，物联网技术被广泛应用于人们的工业生产和日常生活中，为人们的生产和生活提供便利和服务。智能家居(Smart Home)作为物联网技术的研究中的一个典型应用，旨在为人们提供简单、舒适、智能、安全的居住和工作环境，具有广阔的应用前景和商业价值，也是物联网研究工作的重点之一。智能家居的研究主要包括：嵌入式终端和设备、无线传感器网络和异构网络、家居环境的监控与远程控制、体感交互技术(语音识别、手势识别等)、居家安全系统、辅助(智能)决策系统等几个方面。目前智能家居中决策系统的设计过于简单或者完全依赖于人进行。可以通过机器学习的方法来提高辅助决策系统的智能性，减少人在智能家居控制上精力，实现“以人为本”宗旨，提供简单、舒适和智能的环境。本文主要研究BP神经网络算法在智能家居辅助决策系统中的应用。智能家居环境中需要考虑人与人之间的差异，不同的人对环境的要求不尽相同，因此需要考虑不同的人对舒适环境的要求，在进行辅助决策时因人制宜，提高辅助决策的准确性和智能程度。BP神经网络具有良好的自适应和非线性映射能力，可以通过学习掌握智能家居中人的“喜好”，做出最为合适的辅助决策。然而BP神经网络的准确度受神经网络结构和训练样本质量的影响较大，因此需要根据实际环境，合理的设计BP神经网络的结构，优化训练样本的质量，来提升辅助决策系统的准确性。不同的人对环境的要求有差别，导致产生的数据相互影响，降低了BP神经网络训练样本的质量。一种解决方法是通过特殊的方法找出这种差异，对环境中的人根据对环境要求进行分组，然后单独训练出相应的决策规则，辅助决策时对不同的人使用不同的规则，来保证训练样本的质量，提高辅助决策的准确性。本文通过对智能家居中人的行为的描述，结合环境信息，使用K最近邻算法，来表现人对环境要求的差别然后进行分组，确保相同小组的人对环境具有相似的要求，，实现提高决策准确性和智能性的目的。综上所述，本文将神经网络应用于智能家居辅助决策系统中，使用合适的方法来提高样本质量和改进神经网络结构，在提升辅助决策系统的准确性和智能性方面进行了尝试，并通过实验验证了本文方法的可行性，和传统的辅助决策系统比较，具有更好的准确度和智能性，更符合智能家居“以人为本”的思想。
[Abstract]:In recent years, with the continuous development of computer network technology and the cross-application of other technologies, the concept of Internet of things (Internet of things) has been put forward and attached great importance to. At present, Internet of things technology is widely used in people's industrial production and daily life, providing convenience and service for people's production and life. Smart Home, as a typical application of Internet of things technology, aims to provide people with simple, comfortable, intelligent and safe living and working environment. It has broad application prospect and commercial value. Also is one of the focal points of the Internet of things research. The research of smart home mainly includes: embedded terminal and equipment, wireless sensor network and heterogeneous network, monitoring and remote control of home environment, interactive technology of body sensation (speech recognition, gesture recognition, etc.), home security system, etc. Auxiliary (intelligent) decision system, etc. At present, the design of decision-making system in smart home is too simple or completely dependent on people. The method of machine learning can improve the intelligence of the auxiliary decision system, reduce the energy of the intelligent home control, realize the aim of "people-oriented", and provide a simple, comfortable and intelligent environment. This paper mainly studies the application of BP neural network algorithm in intelligent home aided decision system. In the smart home environment, we need to take into account the differences between people, different people have different requirements for the environment, so we need to consider the requirements of different people for comfortable environment. To improve the accuracy and intelligence of auxiliary decision. BP neural network has a good ability of adaptive and nonlinear mapping. It can make the most appropriate auxiliary decision by learning and mastering the "preferences" of people in the intelligent home. However, the accuracy of BP neural network is greatly affected by the neural network structure and the quality of training samples, so it is necessary to design the structure of BP neural network reasonably and optimize the quality of training samples according to the actual environment. To improve the accuracy of the auxiliary decision system. Different people have different requirements for the environment, which leads to the mutual influence of the generated data and reduces the quality of BP neural network training samples. One solution is to find out this difference in a special way, group the people in the environment according to the requirements of the environment, and then train the corresponding decision rules individually, and then use different rules for different people when assisting in the decision. To ensure the quality of training samples, improve the accuracy of decision-making. In this paper, by describing the behavior of people in smart home, combining the environmental information, using K nearest neighbor algorithm, to express the difference of people's environmental requirements and then to group, to ensure that the same group of people have similar requirements for the environment. The purpose of improving the accuracy and intelligence of decision making is realized. To sum up, this paper applies neural network to intelligent home aided decision-making system, using appropriate methods to improve the quality of samples and improve the neural network structure, in order to improve the accuracy and intelligence of the auxiliary decision-making system. The feasibility of this method is verified by experiments. Compared with the traditional auxiliary decision system, it has better accuracy and intelligence, and is more in line with the idea of "people-oriented" in smart home.
【学位授予单位】：吉林大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TU855;TP391.44;TN929.5

【参考文献】