基于自适应集成神经网络的火灾预测方法  

Fire prediction method based on adaptive ensemble neural network

作　　者：李军[1] 张志东 乔元健 高鹤 LI Jun;ZHANG Zhi-dong;QIAO Yuan-jian;GAO He(School of Electronic Information Engineering(University Physics Teaching Department),Qilu University of Technology(Shandong Academy of Sciences),Shandong Jinan 250353,China;School of Electrical Engineering and Automation,Qilu University of Technology(Shandong Academy of Sciences),Shandong Jinan 250353 China;Shandong Zhengchen Technology Co.,Ltd.,Shandong Jinan 250353,China)

机构地区：[1]齐鲁工业大学(山东省科学院)电子信息工程学院,山东济南250353 [2]齐鲁工业大学(山东省科学院)电气工程与自动化学院,山东济南250353 [3]山东正晨科技股份有限公司,山东济南250353

出　　处：《消防科学与技术》2020年第12期1727-1731,共5页Fire Science and Technology

基　　金：山东省研究生教育计划创新项目(SDYY16032);国家级大学生创新创业项目“基于AI多源信息融合火灾预测的智慧消防云控系统关键技术研究”。

摘　　要：针对传统火灾预测方法存在误报和漏报的问题,提出了一种基于自适应集成神经网络的火灾预测方法。首先,在信息层采用速率检测算法将不同类型传感器检测到的奇异数据输入到网络模型中。其次,在特征层采用长短期记忆网络(LSTM)和径向基前馈神经网络(RBF-BPNN)构建集成网络学习不同输入参数下的火灾特征,最后,在决策层设计模糊逻辑控制系统推理输出火灾报警等级。实验结果表明,该方法具有更高的预测精度。Aiming at the problems of false alarm and missing alarm in traditional fire prediction methods, a fire prediction method based on adaptive integrated neural network is proposed. Firstly, the rate detection algorithm is used in the information layer to input the singular data detected by different types of sensors into the network model. Secondly, in the feature layer, the long-term memory network(LSTM) and radial basis function feed forward neural network(RBF-BPNN) are used to build an integrated network to learn the fire characteristics under different input parameters. Finally, the fuzzy logic control system is designed at the decision-making level to infer and output the fire alarm level. Experimental results show that this method has higher prediction accuracy.

关 键 词：自适应集成神经网络 速率检测算法 模糊逻辑控制系统 火灾预测 

分 类 号：X913.4[环境科学与工程—安全科学] TP391[自动化与计算机技术—计算机应用技术]