一种新型数据融合方法在火灾监测中的应用研究  

Application research of a new data fusion approach in fire monitoring

作　　者：李时辉[1] 刘山葆[2] 

机构地区：[1]义乌工商学院机电信息分院,浙江义乌322000 [2]广东省通信管理局,广州510080

出　　处：《江西理工大学学报》2015年第3期85-88,共4页Journal of Jiangxi University of Science and Technology

基　　金：浙江省自然科学基金资助项目(Y1100219)

摘　　要：传统的火灾监测系统一般采用阈值判断法,该系统固然具有快速反应的优点,也存在误报率较高的问题.针对上诉问题,结合火灾预警监测系统中的数据采集要求,在对多组数据处理中,提出了一种基于D-S证据推理的贝叶斯网络法的数据融合新方法.该方法首先利用D-S证据理论推理出贝叶斯网络的信任函数和似然函数的条件概率,然后再求得信度函数和似然函数,最后在决策规则下判断火灾发生的概率.实验结果表明,该方法实现简单,并能有效提高监测系统的判断准确率.The threshold value determination method adopted by conventional fire monitoring system can yield prompt response but it also has a high rate of false monitoring. To solve the problem and to meet the requirements of data collection in fire monitoring system,a new data fusion approach based on Bayesian network algorithm by inference of D-S evidence theory is proposed. The approach uses D-S evidence theory to deduce the conditional probability of Bayesian network＇s belief function and likelihood function and then puts the acquired belief function and likelihood function under decision rule to make judgment on the probability of fire. Experiment result shows that the method is simple and easy to implement, and can effectively improve the accuracy of the monitoring system.

关 键 词：火灾监测 D—S证据理论 贝叶斯网络算法 

分 类 号：TP277[自动化与计算机技术—检测技术与自动化装置]