Research And Implementation Of Intelligent Monitoring System Of Infant Sleep

This thesis will introduce the infant sleep monitoring system, a research project of my laboratory. The system collects various types of sleep data in real-time by IoT(Internet of things) technology, adopts data mining algorithm which is the core of this system to analyze infant behavior and status in real-time, and pushes the alert and result to the users in real-time. As the basic physiological needs of the body, the quality of sleep will directly affect the growth and development of infants. There are a series of questions in current infant sleep monitoring systems in the domestic and foreign market: it works in small areas, it is unable to save the data, it relies on manual too much, the privacy of users cannot be protected and the fee is costly. Therefore, it is imminent to do the research which is to develop a low-cost, high real-time, practical infant sleep monitoring system.The topic of thesis is achieving infant sleep intelligent monitoring system, focusing on the research of support vector machines, but also introducing the overall profile of the project. The concrete structure of the thesis is as follows:The first is the introduction of algorithm used in the infant sleep monitoring system- the theoretical basis of support vector machine algorithm. The thesis focuses on the basic principle of the algorithm including linearly separable problem, linearly non-separable problem and non-linear separable problems, and introduces the part of the practical application of this algorithm.Then the thesis states quadratic programming problem and multi-classification problem which are the difficulties of the algorithm. After making careful compression between different algorithms and their application in the project, we choose the SMO(Sequential Minimal Optimization) algorithm, and one-versus-one algorithm to achieve our goal.The thesis mainly studies how to collect pressure data and how to train and classify the data obtained in the infant sleep monitoring system via support vector machine algorithm. The predicted results for comparison by several experiments verify the feasibility in judging of infant sleep position through the support vector machine algorithm, and also prove the theoretical feasibility of infant sleep monitoring system.In the end, the thesis summarizes the theory of algorithms and system architecture discussed earlier, and systematically introduces the infant sleep intelligent monitoring system whose core is support vector machine. This thesis shows the advantages and the results in the practical application of this sleeping infant intelligent monitoring system and points out its prospects of development on this basis, and summarizes the problems for improvement in the future.