基于方差齐性检验的无线传感网火灾监测

Detection of an emerging fire event by homogeneity test of variance based on wireless sensor network

为了提高火灾监测的精确度和预警的及时性,提出基于无线传感网采集数据的方差齐性检验算法。假设无线传感网中各个簇内数据服从正态分布,根据χ2分布检验算法将各个簇的方差进行假设检验,若各个簇的方差相等,则被监测环境正常;若不完全相等,则表明出现了火灾异常数据,通过F分布的方差检验法将相邻簇的方差排序,依次与最小方差进行比较,找到异常簇,从而实现火灾发生区域的定位,提取出异常簇的温度最大值并报警。仿真实验表明,该算法能够及时反映被监测环境的温度变化,同时能精确定位并根据火灾发生的不同时期给予不同的预警或报警。

In this paper,the algorithm of homogeneity test of variance is proposed to improve the accuracy and timeliness in detection of an emerging fire event.Given that data out of each cluster in wireless sensor network follows normal distribution,the variances of each cluster will be tested by χ2 distribution algorithm,if all of the variances are the same then it implies that the monitored area is in order,otherwise,a fire may be emerging and it brings some abnormal data.The variances of neighboring clusters are ordering by F distribution algorithm,and each of them will be compared with the minimum variance and to find out abnormal clusters.In this way,the area of emerging fire event is located,the maximum temperature of abnormal clusters is Abstracted and a fire warning is started.The simulation results show that the proposed algorithm can timely reveal a change of temperature of the monitored area,in the meantime,it can accurately locate the area of emerging fire event,and a systematic pre-warning or warning will be sent according to the discriminative stages of emerging fire event.