【作者】 张博；

【导师】 张洪欣；

【作者基本信息】 北京邮电大学 ， 生物医学工程， 2016， 硕士

【摘要】 本论文研究的对象为通信信号和生物医学信号,研究的主要内容为通信信号的特征提取和调制方式识别,以及基于传感器的生物医学信号特征提取和聚类分析。论文中的通信信号处理部分是依托于实验室的项目,在现代信号处理算法的基础上,综合运用统计学知识和机器学习算法,实现多种典型通信信号调制方式的识别和信噪比的估算。针对通信信号,采用传统信号处理理论和机器学习技术相结合的方法,对信号进行特征提取,使用时频域结合分析的方法在决策树理论基础上实现七种典型通信信号调制方式的识别。在实现的过程中笔者突破传统的单一特征提取方法,通过总结和研究不同调制方式具有的不同特点,然后根据这些特点在第一次特征提取的基础上做二次特征提取,即提取完特征后对特征向量再进行一次特征提取,有效地提高了识别率和识别算法的稳定性。在研究单一信号调制方式识别的基础上,进行了混合信号的分离及识别,使用带通滤波器将混合信号分离成单独的信号,然后按照单一信号的识别进行分析,目前仅能实现最多两种信号的混合分离及识别。在研究通信信号调制方式识别的过程中,需要更好地排除环境干扰和噪声干扰,笔者研究了一些常见干扰信号的识别,包括高斯信号、窄带脉冲信号和变压器电源辐射出来的信号。这些种类的信号与调制信号在信号产生原理上存在较大的不同,仅通过对时域和频域结合的方法就能将其识别出来。本文中通信信号调制方式识别过程中的信号接收设备为示波器,其优点是通用性强,易于与笔记本电脑进行仪器通信和交互。除了识别典型信号的调制方式外,还完成了六种典型信号的信噪比估计。按照信号的包络特征,将信号分为恒包络信号和非恒包络信号,分别使用功率比对法和奇异值分解法估算出信噪比,在0~18dB范围内,估算偏差不超过0.5dB。生物医学信号方面,采用的是数据挖掘和分析技术。本文研究了一个目前在国内来说较为新颖的领域,充分利用智能手机的各项传感器功能,将手机作为信号采集设备,同时用作信号处理工具,不借助任何医学设备,实现随时随地测量多个医学指标的功能。目前笔者开发出一款手机APP,能够实现对心率的测量,对血压的测量还停留在仿真阶段,后续需要投入更多的时间进行更深入的研究。在这个关注健康且智能手机盛行的时代,笔者坚信这样的研究是有意义的,并且在未来会有更多的人来开拓这方面的研究。更多还原

【Abstract】 Research object of this paper are communication signals and biomedical signals,the main content of the study for communication signal are feature extraction and modulation recognition,and for biomedical signal are feature extraction and cluster analysis,based on sensors.For communication signals,features are extracted,using the combination of traditional signal processing theory and artificial neural network method,and using the combined method of time domain and frequency domain based on the theory of decision tree to identify seven modulation signals.With summarizing the characteristics of various modulation method,the author breaks the traditional feature extraction methods,strengthens the features,and extracts features for the second time.This method effectively improves the recognition rate and algorithm robustness.During the process of implementation,the author broke through traditional single feature extraction methods,by summing up and studying different modulation mode.according to these characteristics,and on the basis of feature extraction for the first time,I do secondary feature extraction,namely after extracting the feature to extract feature vector again,effectively improving the recognition rate and recognition algorithm of stability.In the study of the single signal modulation mode recognition,on the basis of the separation and identification of mixed signal,I used the band-pass filter to separate mixed signal into separate signals,and then according to the identification of a single signal.Currently it can only achieve a mixture of at most two signal separation and identification.In the process of the communication signal modulation mode identification,it needs to eliminate interference and noise,the author studied some common jamming signal recognition,including gaussian signal,narrow pulse signal and radiate transformer power supply of the signal.These kinds of signals and on the principle of signal modulation signal is different,only through the combined method of time domain and frequency domain can be identified.This article communication signals modulation mode identification in the process of receiving equipment for the oscilloscope,its advantage is strong commonality,easy to instrument communication and interaction with laptop computers.According to the signals’ characteristics of envelope,signals can be divided into constant envelope signals and non-constant envelope signals,respectively using power ratio method and(SVD)singular value decomposition method.In this paper,the author completes six kinds of signals’ SNR estimation by simulation.In terms of biomedical signals,only based on smartphones,data mining and analysis techniques are used to measure many kinds of indicators of medicine anytime and anywhere,without any medical equipment.At present,I developed a mobile phone APP,which can measure heart rate.And the measurement on blood pressure still stays in the simulation stage,the topic need more time for further research.With the increasing focus on health and the popularity of the smartphones,I believe that such a study is meaningful and there will be more people to develop in this field in the future.更多还原