摘要
传统的室内火源定位方法一般都是针对无风条件下的火源进行定位,但实际火灾现场往往是有风的。为了研究有风环境下火源点的定位问题,提出一种基于数据驱动的火灾定位方法,结合极限学习机优秀的训练速度和粒子滤波对于非线性系统的优越性,拟合得到出有风环境时的火灾温度场系统量测方程,结合粒子滤波进行定位,并与常用的最小二乘法定位结果进行对比,仿真结果表明本文方法在精确性与稳定性上优于最小二乘法。
Traditional indoor fire source positioning method is generally aimed at the fire source under ideal conditions and the model is relatively simple.But in fact,the scene of the fire is often with wind.In order to study the location problem of fire source point in windy environment.A method base on data-driven extreme learning machine(ELM) combined with particle filter(PF) is proposed,which combines the excellent training speed of extreme learning machine and the superiority of particle filter for nonlinear systems.The system of measurement of fire temperature field system in windy environment is obtained,and the particle filter is used to locate and compare with the least squares positioning result.The simulation results show that the proposed method is better than the least squares in accuracy and stability.
作者
闫东
冯肖亮
王国勇
YAN Dong;FENG Xiaoliang;WANG Guoyong(College of Electrical Engineering,Henan University of Technology,Zhengzhou Henan 450000,China;School of Computer and Information Engineering,Luoyang Institute of Science and Technology,Luoyang Henan 471023,China)
出处
《杭州电子科技大学学报(自然科学版)》
2019年第1期75-80,共6页
Journal of Hangzhou Dianzi University:Natural Sciences
基金
国家自然科学基金资助项目(61503174)
河南省科技攻关资助项目(162102210196)
河南省教育厅自然科学研究资助项目(15A413011)
关键词
火源定位
非线性建模
极限学习机
粒子滤波
fire source location
nonlinear modeling
extreme learning machine
particle filter