机器人主动嗅觉烟羽分布辨识方法研究

An Identification Method of Plume Distribution Using a Robot with Active Olfactory

针对三维空间有害气体烟羽分布的辨识问题,提出一种模拟退火结合模糊C均值算法的烟羽辨识方法。采用模拟退火算法优化四轴飞行器自主搜索路径,避免了烟羽信息采集的盲目性,提高了辨识效率。采用模糊C均值算法对有害气体烟羽多维特征数据聚类,辨识烟羽扩散区域及危害程度。采用两种聚类评价指标,通过实验确定了聚类数目及加权指数。搭建了仿真平台,通过实例对所提出的方法进行验证,且与Kmeans、DBSCAN聚类算法对比分析,结果表明:上述方法能自主辨识出有害气体烟羽扩散区域及危害程度,为烟羽分布辨识及源定位提供方法支持。

In order to identify the distribution of the harmful gas plumes in three-dimensional space,a plume identification method is proposed based on simulated annealing combined with fuzzy C-means algorithm.Simulated annealing was used to optimize the autonomous search paths of a four-axis aircraft to avoid the searching blindness.The fuzzy C-means algorithm was used to cluster the multi-characteristic data of the harmful plumes to identify the plumes diffusion areas and the damage degree.According to two clustering evaluation indexes,the number of clusters and the parameters of the clustering algorithm were determined by experiments.A simulation platform was set up to verify the proposed method via illustrations.The results show that this method can identify the plumes diffusion areas and the harmful gas damage degree,and provide a method support for the plumes distribution identification and the location of plume sources.