近年来，随着人们生活水平的提高和居住条件的改善，大量的新型装饰材料被广泛地应用于家庭装修中，而由此所造成的室内环境污染也日益变得严重。那些有毒气体会对人体造成极大的危害，为了保障人们的身体健康，发展高性能、低成本的多传感器与智能技术相结合的监测系统具有重要的现实意义。基于多传感器数据融合技术的气体检测系统，可在一定程度上抑制传感器的漂移和噪声，提高系统的检测精度。本文采用基于多传感器自适应加权数据融合技术，设计一套室内装修有毒气体监测系统，该系统具有数据采集、实时显示、声光报警、串口通讯等功能，并采用自适应加权数据融合算法，提高了整个系统的测量精度。 论文在分析了电化学甲醛、氨气传感器和半导体苯传感器的检测原理之后，提出并实现了系统整体设计方案。系统硬件主要分为3个部分：一是传感器信号调理电路部分，包括甲醛、氨气传感器的信号调理、苯传感器的信号调理和温湿度传感器的信号调理；二是TMS320LF2407A控制器部分的电路，包括气体浓度的采集和温湿度参数的采集、液晶显示电路、声光报警等；三是上位机通讯电路。系统软件包括两个部分：一是用C和汇编语言混合编写的下位机程序，主要完成系统初始化、中断采样、数据实时显示、声光报警和数据通讯等工作；二是由VB语言编写上位机软件，主要完成数据实时传输、数据保存、显示气体浓度等功能。 论文提出了基于自适应加权的数据融合方法，即充分利用传感器的原始数据，对测量数据的均方误差、测量精度等信息进行融合，并准确估算出来。其优点是不要求测量数据的任何先验知识，仅依据测量数据的测量精度来确定不同数据的相应权数，就可以计算出均方误差最小的融合值。将所提出的数据融合方法应用到室内有毒气体监测系统中，提高了室内有毒气体监测系统的测量精度，并使系统具有良好的鲁棒性。 实验表明系统性能稳定，测量精度达到预定的要求，具有很好的应用前景。

With the improvements of the standard of living and the condition of housing in recent years, a lot of new-style decorating materials are widely used in house decoration, which caused the indoor environment deteriorated. Those poisonous gases have caused great harm to human health. So it is of great importance to develop the high quality and low cost monitoring system by combining multi-sensor and intelligence technology. Multi-sensor data fusion technology based on gas detecting system can restrain the excursion and the noise of the sensor and improve the accuracy of the system. Based on the technology mentioned above, the thesis devises an indoor decorating poisonous gas detecting system. The system comprises the data real-time displaying part, the acousto-optic alarm part and the serial port communication part. In addition, the data fusion method based on adaptive weighted is used in the system to improve the accuracy of the system. The thesis deeply analyzes the detection principles of electrochemical formaldehyde, ammonia sensor and semiconductor toluene sensors. Then a system plan is put forward and is implemented. So the analyses for software and hardware design of the system are given. The hardware section consists mainly of three parts. The first part is a series of sensor signal conditioning circuits, including formaldehyde and ammonia sensor conditioning circuit, formaldehyde sensor signal conditioning circuit and temperature- humidity sensor signal conditioning circuit. The second part is the TMS320LF2407A Controller circuit including the gas concentrations acquisition, the temperature- humidity parameters acquisition, the liquid crystal displaying circuit and the acousto-optic alarm and so on. The third part is a position machine communication circuit. The software section consists of two parts. One is the program with C language and assembles language embedding machine. In this part, the performances of the system initialization, the interrupt sampling, the data real-time displaying, the acousto-optic alarm and the data communications are realized. The other is the position machine software based on VB language. It mainly achieves the real-time transmission, the data reservation and the displaying gas concentration. The data fusion method based on adaptive weighted is presented in this thesis. Namely, the mean square error (MSE) and the measurement accuracy of the raw data from these multi-sensors are fused and are estimated accurately. The advantage is no need the prior knowledge of data measurement but only according to the measurement accuracy when we make sure the corresponding weights of the different data. In the light of this method we can calculate and gained the minimum fusion value and then apply it to indoor poisonous gas detecting system to improve the accuracy and robust of the system. In the end, the real experiments show that the system has a steady quality and high accuracy as required. The system is of a bright future.