基于改进D-S证据理论的变电站火灾监测报警技术研究

Research on Substation Fire Monitoring and Alarm Technology Based on Improved D-S Evidence Theory

为了解决传统火灾监测预警算法采用单一传感器模式对火灾信息评估不完整,容易出现误判、漏判的问题,基于改进的D-S证据理论,构建了基于多传感器信息融合算法的变电站火灾报警系统。得出如下主要结论:常规D-S证据理论在信息融合过程中容易出现一票否决、Zadeh悖论以及公平性问题,通过引入相似度概念对基本信度分配函数进行修正,可有效解决这些问题;利用改进的D-S证据理论对多传感器信息源进行融合处理,同时将隶属度函数Sigmf作为变电站火灾监测传感器的基本概率函数,可对火灾不同阶段的特征进行监测和提取;当采用多个传感器数据进行融合处理后,增强了信息的冗余性和互补性,减少火灾误判和漏判的概率,使得变电站火灾预警准确率得到有效提升。

In order to solve the problem that the traditional fire monitoring and early warning algorithm uses a single sensor mode to evaluate the fire information incompletely,which is prone to misjudgment and missed judgment,based on the improved D-S evidence theory,a substation fire alarm system based on multi-legend information fusion algorithm is constructed.The main conclusions are as follows.In the process of information fusion,conventional D-S evidence theory is prone to one vote veto,Zadeh paradox and fairness problems,which can be effectively solved by introducing the concept of similarity to modify the basic reliability allocation function.The improved D-S evidence theory is used to fuse multi-sensor information sources,and the membership function Sigmf is used as the basic probability function of substation fire monitoring sensor,which can monitor and extract the characteristics of different stages of fire.When multi-sensor data are fused,the redundancy and complementarity of information are enhanced,the probability of misjudgment and missed judgment is reduced,and the accuracy of substation fire early warning is effectively improved.