【作者】 王振军；

【导师】 张文超；

【作者基本信息】 杭州电子科技大学 ， 电路与系统， 2010， 硕士

【摘要】 随着传感器技术、电子技术、计算机技术和模式识别技术的发展,电子鼻的应用领域越来越广泛,受到更普遍的关注。电子鼻的应用领域已经渗透到包括质量控制、环境保护、国家安全、消防、汽车、航空航天、疾病诊断和工业过程控制等领域。研究实现低成本、高精度、高集成度、商品化的电子鼻系统可以弥补传统气体检测方法(如气相色谱法)操作复杂、体积大、便携性差、成本高、无法实时自动监测等缺点。论文在分析研究电子鼻系统的原理和特点的基础上,根据BP网络在电子鼻系统中的应用原理,提出了将BP网络模式识别用于电子鼻系统的总体技术方案。系统采用MSP430微处理器作为下位机控制器完成A/D转换、数据存储、RS232串口通信、LCD显示和人际交互,采用PC机作为上位机完成数据采集、模式识别功能。在总体方案的基础上设计了E2PROM、RS232、LCD显示、系统电源、键盘等硬件电路及其软件,能够在MSP430控制器的控制下各部分协同工作,并给出了硬件电路和软件流程。考虑到气体传感器的功耗较大,论文设计了输出电压恒定为5V、最大输出功率为15W的传感器阵列加热控制电路。根据传感器阵列的输出电压信号变化范围,设计了传感器阵列的信号测量、放大及A/D转换电路。根据BP网络应用特点,设计了电子鼻上位机软件和BP网络模式识别算法,并给出了系统的功能测试结果。研制的电子鼻系统进行了单一气体定性分析和定量计算实验,实验结果表明BP网络模式识别算法用于电子鼻系统是切实可行的,并取得预定效果。本文还对影响BP网络训练和网络性能的关键因素进行了研究。此外,本系统对于采用传感器阵列和BP网络模式识别算法的电子鼻系统具有一定的通用性。更多还原

【Abstract】 With the development of sensor, electronics, computer and pattern recognition technology, the application fields of electronic nose systems are being more and more widespread and also receiving a more universal attention. Electronic nose’s applications have penetrated into quality control, environmental protection, national security, fire control, automobile, aerospace, disease diagnostics and industrial process control and other fields. Research and implementation of electronic nose system of low-cost, high accuracy, high integration and commercialization can make up the shortcomings of operation complicated, bulky, poor portability, high cost and no real-time automatic monitoring of traditional gas detection methods (such as gas chromatography).With analyzing the principles and characteristics of electronic nose system, the paper puts forward the overall technical solution of electronic nose system using BP neural network as the pattern recognition. According to the theory of BP neural network used in electronic nose systems, system uses the MSP430 microprocessor controller as subordinate computer to complete A/D conversion, data storage, RS232 serial communication, LCD display, and interpersonal interaction, and also uses PC as host computer to complete data acquisition, pattern recognition. On the basis of the overall solution, paper also designs hardware circuits and the software of E2PROM, RS232, LCD display, system power, keyboard and others, all these parts can work together under the control of MSP430, and gives realization of the hardware circuits and program flow of the software. Considering the larger power consumption of gas sensors array, paper designs the heating control circuits with constant 5V and 15W maximum output power. And paper also designs signal measurement, amplification and analog-digital conversion circuits. According to the characteristics of electronic nose using BP network, paper designs host computer software and BP network pattern recognition algorithms, and gives the system’s functional test’s results.At the end of the paper, the experimental results of single gas’s qualitative analysis and quantitative calculation using developed system shows that BP network pattern recognition algorithm applied to electronic nose system is feasible, and able to achieve the desired results. Paper also researches key factors which affect the performance and train of BP network. In addition, the system has a certain degree of generality for electronic nose system which using sensor arrays and BP network pattern recognition algorithm.更多还原