Forest Fire Monitoring System Based On Multi-sensor Data Fusion

Forest fire monitoring system has now become a major monitoring way to prevent fire and protect the forest by countries on a global scale.At present, besides conducting studies on conventional methods, these countries also began to experiment on applying wireless sensor networks to forest fire monitoring system. This new technique will help people effectively and quickly detect potential fire hazard and make corresponding coping measures.Compared with the massive sources of manpower and material consumed under the conventional monitoring mode, wireless sensor networks has advantages of wide coverage, energy efficiency and low cost. It can better perform monitoring to protect ecological environment and people’s property. Currently, how to better apply wireless sensor networks int to forest fire monitoring system to improve the accuracy of system monitoring has gradually become a valuable research direction.Forest fire monitoring system based on wireless sensor network technology collected the concerned parameters (such as temperature, humidity, etc.),through the sensors, and then sent the data to the computer of monitoring center, the computer of monitoring center made fusion analysis processing according to the data collected, got conclusion through the judgment decision. There are many questions In the existing forest fire monitoring system based on wireless sensor network, one of these is the data processing field. This paper put forward a gradient data fusion model for solving the several problems such as the high computational complexity an, low value accuracy due to the large amount of data,and only judge the results of a single signal, multiple signals together to judge.After investigating the application of wireless sensor networks into forest fire monitoring system, this article analysed the strengths and weaknesses of various data fusion algorithms, especially that of correlation function, least square method, and DS evidence theory, etc.Based on problems such as low accuracy of fusion value in forest fire monitoring system, it introduced a multi-sensor data fusion mode. Firstly, design and set up an intermediate staging point between the sensor and the fusion center, and get the mutual support between each sensors using data collected by the sensor based on an algorithm for correlation function. Secondly, delete data from sensors with low support, and use the method of least square to fuse date from those with high support. At last, put the previous fused data into a global fusion by using a improved DS evidence theory. The theoretical analysis result proved that the precision of the multi-sensor data fusion mode is higher than the conventional data fusion mode. Using information shared by Meteorological Bureau of Taiwan Ministry of Communications and database identified by academic circles, with two modes fusion value getting from Matlab simulation platform, these two modes Fusion value are converted into probabilistic contrast, which proved the idea that multi-sensor data fusion mode has higher precision and effectively improves judgement’s reliability after optimization raised by this article.