论文针对我国煤矿井下局部通风机长期恒速运转，“一风吹”比较严重的现象，将数据融合理论应用到瓦斯智能排放系统中，设计了一种基于数据融合理论的瓦斯智能排放系统，节约了电能，有效地预防了瓦斯事故的发生。 传统的数据融合方法一般分为两大类：基于时间和基于空间的多传感器数据融合方法。这些方法都有效地提高了测量精度，但由于割裂了数据融合的时间性和空间性，所以这些方法都具有一定的局限性。论文利用时空融合估计算法，先将每个传感器在不同时刻的观测值与该时刻之前的测量初值进行融合，得出该传感器在不同时刻的融合估计值，然后将各个传感器同时刻的估计值进行空间融合，提高了测量精度。 传统的瓦斯排放系统主要考虑的是瓦斯浓度，本文将数据融合理论应用到瓦斯排放系统中，除了考虑瓦斯浓度，还考虑了CO、温度和粉尘对人的影响，更加全面地考虑了煤矿井下的环境。论文首先对传感器进行基于算术平均递推估计的时间融合，然后对传感器进行基于自适应加权数据融合算法的空间融合，提高了测量精度。然后利用BP神经网络，以瓦斯，CO，温度，粉尘等传感器的数据为输入，以风速为输出，结合Matlab仿真技术，成功地完成了对煤矿井下通风风速的预测。论文对风速模糊控制器进行了设计，控制器以掘进巷道风速的偏差和偏差变化率作为输入，控制变频器的输入电压为其输出。论文将数据融合和模糊控制思想应用到瓦斯智能排放系统中，在一定程度上提高了采煤工作面瓦斯智能排放系统的可靠性，有效地提高了系统的智能化水平和安全性指标。

In view of the phenomenon of the local fan working in the long-term constant speed and the invariable speed, data fusion theory is applied in drainage system of gas and an intelligent drainage system of gas based on data fusion theory is designed for local fan. It can not only save energies but also be easy to prevent the gas accident events. The traditional methods of data fusion have two different categories: the one is data fusion method based on time and the other is data fusion method based on space. The measurement precision is enhanced by using the two methods effectively. Because the time and space are separated, these methods have their limitations. The estimation algorithm based on time-space is used in this paper. In this algorithm, the data that are monitored by the same sensor at the different time are fused to gain the estimated data about the sensor at the different time. Then the data that are fused in time are fused in space. The measurement precision is enhanced. The gas is mainly considered in the traditional intelligent drainage system of gas. The gas is not only considered but also the harm of CO ,the temperature and the dust is considered in this paper. More comprehensive factors of environment are considered in the mine. Firstly, the recursive fusion estimation based on time and adaptive weighted fusion algorithm based on space are used to improve the measurement precision. Then BP network and Matlab simulation are used to predict the speed of the local fan. In the model of Matlab simulation, input variables are data from gas sensors, carbon monoxide sensors, temperature sensors, coal dust sensors. Output variable is data of wind speed . Then fuzzy control algorithm is used, whose input variables are gas density error and gas density error change and output variable is the control voltage of the transducer. The reliability and automation grade of the system are highly raised in virtue of fuzzy control theory and data fusion theory. The intelligence and safety of the local fan gas drainage system are aslo enhanced in virtue of fuzzy control theory and data fusion theory.