面向气体传感器的自检测智能算法与硬件系统研究
发布时间:2018-05-05 08:17
本文选题:大气检测系统 + DSP ; 参考:《吉林大学》2017年硕士论文
【摘要】:近年来愈演愈烈的雾霾天气使人类意识到大气环境质量保护的重要性。环境质量的改善是一项长期的工作,需要对环境进行调查、检测后分析、制定治理方案。环境监测需要借助先进的监测技术和监测仪器,检测结果直接影响到环境治理方案的制定。因此,可靠、准确地大气环境监测对大气环境的治理和保护至关重要。研究基于气体传感器阵列的智能化大气检测系统实现对大气环境的准确监测。该系统的研究主要从三个方面进行:系统硬件电路的设计、GUI(图形用户界面,Graphical User Interface)人机交互界面的制作和自检测智能算法的研究。系统硬件主要包括气体传感器阵列的驱动电路、信号调理和数据采集电路、信息处理单元电路、进气部分控制电路、外部接口电路。主控制器使用DSP2000系列的F28335,该芯片既能实现对外围电路的控制。由气体传感器阵列的驱动电路、信号调理和数据采集电路、进气部分控制电路和外部接口电路构成了数字信号采集电路,为智能算法处理和人机交互界面显示提供了原始数据,是数据流动的源头。人机交互界面放弃常用的液晶屏显示方式,基于QT制作类似手机、平板界面一样更真实、美观的操作界面,完成系统功能高度集成,实现傻瓜式操作。使用QML渲染界面,建立实现界面操作行为的C++功能函数库,这种非阻塞式设计满足前端与后端的同时运行。用户通过操作界面触发信号控制C++函数,与主控制器F28335交换数据和控制指令,实现硬件电路控制。系统采用主成分分析(PCA:Principal Component Analysis)、支持向量机(SVM:Support Vector Machine)和支持向量回归(SVR:Support Vector Regression)三种算法结合,对大气气体鉴别和浓度获取。使用主成分分析法(PCA)对气体传感器采集数据进行预处理,通过降低维度来滤除混入的信息,从而实现降低信号的噪声。支持向量机(SVM)是一种学习型分类器,通过训练数据集建立样品分类的模型和关于浓度的支持向量回归模型,使用测试数据检验模型建立的效果。使用MATLAB调试自检测智能算法,实现数据的处理和结果图像的显示。根据传感器的结构特点和专家经验,通过全面展开传感器失效机理的研究,进行深入的故障模式的剖析分解,使用支持向量回归算法建立模型,实现故障的诊断和恢复。检测系统各项功能正常运行,以NO_2、SO_2、O_3、CO四种气体为检测对象,设计四组实验获取数据,使用MATLAB图像和人机交互界面两种方式显示处理结果。实验结果表明系统在气体鉴别和浓度测试方面具有较高的准确度,并且操作简单、成本低廉,具有良好的应用前景。
[Abstract]:In recent years, more and more haze weather has made people realize the importance of environmental quality protection. The improvement of environmental quality is a long-term task. Environmental monitoring needs advanced monitoring technology and monitoring instruments, and the results directly affect the formulation of environmental control programs. Therefore, reliable and accurate monitoring of atmospheric environment is very important to the management and protection of atmospheric environment. An intelligent atmospheric detection system based on gas sensor array is studied to realize accurate monitoring of atmospheric environment. The research of the system is mainly carried out from three aspects: the design of the hardware circuit of the system and the design of the graphical User Interface (HMI) and the research of the intelligent algorithm of self-detection. The hardware of the system mainly includes the driving circuit of gas sensor array, signal conditioning and data acquisition circuit, information processing unit circuit, air intake control circuit, external interface circuit. The main controller uses DSP2000 series F28335, which can control the peripheral circuit. The digital signal acquisition circuit is composed of driving circuit of gas sensor array, signal conditioning and data acquisition circuit, air intake control circuit and external interface circuit, which provides raw data for intelligent algorithm processing and man-machine interface display. Is the source of data flow. The man-machine interactive interface gives up the usual LCD display mode, and makes the similar mobile phone based on QT. The flat interface is more realistic and beautiful. The system functions are highly integrated and the fool type operation is realized. Using QML rendering interface, the C function library is established to realize the interface operation behavior. This non-blocking design can meet the needs of the front-end and back-end simultaneously. The user triggers the signal control C function through the operation interface, exchanges data and control instructions with the main controller F28335, and realizes the hardware circuit control. The system adopts three algorithms: principal component analysis (PCA), support vector machine (SVM) and SVR: support Vector (SVR) to identify the atmospheric gas and obtain the concentration. The principal component analysis (PCA) is used to preprocess the data collected by the gas sensor, and to filter out the mixed information by reducing the dimension, so as to reduce the noise of the signal. Support Vector Machine (SVM) is a kind of learning classifier. The model of sample classification and the support vector regression model about concentration are established by training data set, and the effect of the model is verified by test data. The intelligent algorithm of self-detection is debugged by MATLAB to realize the data processing and the display of the result image. According to the structural characteristics and expert experience of the sensor, the failure mechanism of the sensor is studied in an all-round way, the fault mode is analyzed and decomposed deeply, and the support vector regression algorithm is used to establish the model to realize the fault diagnosis and recovery. The functions of the detection system are running normally. Four groups of experiments are designed to obtain the data, and the processing results are displayed by MATLAB images and man-machine interface. The experimental results show that the system has high accuracy in gas identification and concentration measurement, simple operation, low cost and good application prospect.
【学位授予单位】:吉林大学
【学位级别】:硕士
【学位授予年份】:2017
【分类号】:TP18;TP212
【参考文献】
相关期刊论文 前6条
1 梅笑冬;王彪;朱哲;赵培陆;胡小龙;卢革宇;;基于支持向量机的香水识别电子鼻系统设计[J];吉林大学学报(信息科学版);2014年04期
2 孙俊;王艳;金夏明;毛罕平;;基于MSCPSO混合核SVM参数优化的生菜品质检测[J];农业机械学报;2013年09期
3 赵发定;周正义;穆帅欢;阳路;;DSP和AD9244的多路数据采集系统[J];微型机与应用;2013年17期
4 赵万;梅笑冬;朱哲;张厚博;胡小龙;王彪;卢革宇;;基于Labview和PCA分析方法的电子鼻设计[J];吉林大学学报(信息科学版);2013年02期
5 解庆;杨武;赵小强;;基于KPCA与LS-SVM的化工过程故障诊断算法研究[J];工业仪表与自动化装置;2012年05期
6 金向军;张勇;谢云飞;丛茜;赵冰;;基于支持向量机及小波变换的人参红外光谱分析[J];光谱学与光谱分析;2009年03期
,本文编号:1846931
本文链接:https://www.wllwen.com/kejilunwen/zidonghuakongzhilunwen/1846931.html