基于K-means算法的非均匀网格化空间采样分布优化

K-means Algorithm Based on Non-uniform Mesh Spatial Sampling Distribution Optimization

电磁环境地图构建中所用到的空间采样方法多种多样。传统的均匀网格化采样,数据点的选取比较规则,均匀分布在各个传感器节点,方便电磁地图的构建。当区域大量存在有规律的空间分布模式时,采用此采样方式则会得出片面的结果,我们提出了一种非均匀网格化采样分布,该方法是在具有一定相似特征的区域内选取一个具有代表性的数据点,区域的形状大小不统一规定,而是根据各个数据点的特征将其聚类。聚成多类之后,选取各个类的中心点作为数据采样点。经过仿真验证和实地数据测量,可以看出数据点的选取符合非均匀网格化采样,证明了该方法的可行性和有效性。

In the process of the construction of the electromagnetic environment map,various sampling methods are used.For the traditional uniform grid sampling,the selection of data points is rules,evenly distributed in each node. But when the spatial distribution of the regional landscape abundant regular pattern,using the sampling will be one-sided results,so we put forward a non-uniform grid sampling distribution,characteristics of this method is similar in a certain area to select a representative data points,the shape of the area size is not uniform,according to the features of each data point to the cluster.After clustering,select the center points of each class as data sampling points. Through simulation and field measurement data,we can see that the selection of data points in non-uniform grid sampling,and proves the feasibility and veracity.