全量程氢气检测微系统集成设计与信号处理方法研究

This project is served for the widely actual demands of the aerospace industry and the hydrogen producing process. Combined with MEMS technology and sensor signal processing methods, this project focuses on the full-range hydrogen detection microsystem and achieves the online full-range hydrogen concentration measurement with no gap. Miniaturization, integration and intelligent are the key scientific issues in sensor technology development. The micro-structure hydrogen sensor developed in this project gets rid of the traditional craft conception. Instead, the sensor artfully integrates three kind of sensing units with the principles of semiconductor, catalytic combustion, thermal conduction, respectively. Also, based on the optimal structure design, the sensor chip is manufactured by the MEMS technology. The approaches to measuring the full-range hydrogen concentration based on ternary signal processing and small sample calibration data are researched in depth in both theory and practice. Besides, the mathematics model of the relationship between hydrogen concentration and ternary sensor output data is built, which can improve the accuracy of hydrogen detection. What is more, the theories of sparse kernel principle component analysis (SKPCA) and fuzzy relevance vector machine (FRVM) are applied to this project for the purposes of achieving sensor self-diagnosis as well as indicating the uncertainty of hydrogen concentration measurement. At the same time, the sensor will have the function of self-validating and be more intelligent. Finally, the completive algorithms will be solidified in the hardware with the sensitive units, and then a micro-system is obtained. The micro-system is so promising that it can be widely used for hydrogen detection in energy, aerospace and other industrial fields. The creative achievements of this project will promote the development of integrated sensor and smart sensor with a broad application prospect and utility value.

本项目面向航空航天、制氢工业生产过程的广泛实际需求，结合MEMS技术和传感器信号处理方法，研究全量程氢气检测微系统，实现氢气浓度的全量程无缝在线测量。集成化、微型化和智能化是传感器技术发展的关键科学问题。拟研制的微结构氢气传感器突破传统工艺概念，巧妙集成半导体、催化燃烧、热导三种不同原理敏感单元，在优化设计结构的基础上,以MEMS工艺构造敏感芯片。在理论和实践上深入研究三值性信号处理和小样本标定数据下氢气浓度全量程测量方法，建立氢气浓度与三值输出的数学模型，提高氢气检测精度。引入稀疏核主元分析和模糊相关向量机理论，实现传感器故障自诊断，输出氢气浓度不确定度信息，使传感器具有自确认功能，智能化程度得到有效提升。研究的算法成熟后将固化在硬件上与敏感芯片封装在一起组成微系统，广泛应用于能源、航天等工业领域的氢气检测。项目的创新性成果将推动集成化、智能化传感器的发展，具有广阔的应用前景和实用价值。

集成化、微型化和智能化是传感器技术发展的关键科学问题。本项目面向航空航天、制氢工业生产过程的广泛实际需求，结合MEMS 技术和传感器信号处理方法，研究全量程氢气检测微系统，实现了氢气浓度的全量程无缝在线测量。研制的微结构氢气传感器突破传统工艺概念，巧妙集成半导体、催化燃烧、热导三种不同原理敏感单元，在优化设计结构的基础上，以MEMS工艺构造了敏感芯片。建立了氢气浓度与三种不同类型敏感单元输出信号的数学模型，在小样本标定数据下利用相关向量回归模型实现了氢气浓度的全量程无缝精确测量，提高了氢气检测精度。提出基于核主元分析的氢传感器在线检测方法，基于相关向量回归的氢传感器故障数据恢复方法，以及核主元分析特征提取和相关向量回归多分类器的氢传感器故障诊断方法，提高了氢传感器系统的可靠性。在此基础之上，提出一种B类不确定度计算方法，建立了氢气浓度测量值不确定度的传递模型，实现了自确认功能。最后，设计和制作了全量程氢气检测微系统硬件平台。系统硬件体系结构包括样气检测气室，信号调理电路和数据采集电路，DSP 及其外围电路、电源接口电路等。以DSP为核心，由DSP控制A/D转换，实现各种算法获取氢气浓度信息和各敏感单元状态信息。通过实验分析，对系统的各项性能指标进行综合测试。研究的算法可固化在硬件上与敏感芯片封装在一起组成微系统，广泛应用于能源、航天等工业领域的氢气检测。项目的创新性成果将推动集成化、智能化传感器的发展，具有重要的应用前景和实用价值。