基于SVM的火灾警报系统传感器组合优化研究  

Research on sensor combination optimization of fire alarm system based on SVM

作　　者：朱江 徐梦瑶 李达 宋大成 高福海 ZHU Jiang;XU Mengyao;LI Da;SONG Dacheng;GAO Fuhai(School of Air Transportion,Shanghai University of Engineering Science,Shanghai 201620,China)

机构地区：[1]上海工程技术大学航空运输学院,上海201620

出　　处：《智能计算机与应用》2020年第3期313-317,共5页Intelligent Computer and Applications

摘　　要：为了提高对火灾的监控,提高火灾警报系统的灵敏性与可靠性,本文将温度传感器、烟雾传感器、一氧化碳传感器、二氧化碳传感器和氧气传感器的数据进行处理转换,作为SVM算法的输入,利用SVM算法的分类泛化能力对火灾进行精准的识别,最后求得最佳性能的探测器类型的组合。在国内尚未对各类常用传感器的组合进行评估的情况下,本文全面而精准地测算了上述五类传感器各种组合在实验中的优劣,并且得出在精确度足够高的情况下,成本最低最具性价比的传感器组合为:温度传感器、烟雾传感器和一氧化碳传感器。In order to improve the monitoring of fire disaster and improve the sensitivity and reliability of fire alarm system,the data of temperature sensor,smoke sensor,carbon monoxide sensor,carbon dioxide sensor and oxygen sensor are processed and converted,which is used as input data of SVM algorithm.The classification generalization ability of the SVM algorithm accurately identifies the fire disaster,and obtains the combination of the detector types with the best performance based on the experimental results.This paper accurately measures the advantages and disadvantages of the various combinations of the above five types of sensors in the experiment,and finds that in the case of high accuracy,the lowest and most cost-effective sensor combination is:temperature sensor,smoke sensor and carbon monoxide sensor.

关 键 词：火灾 探测器 组合 SVM 

分 类 号：TP212[自动化与计算机技术—检测技术与自动化装置]