Abstract

For industrial logistics tasks, an algorithm for clustering a given set of machine parts on the basis of engineering drawings is proposed. To speed up clustering, the values of each parameter are fuzzified and local maximums of the N-dimensional histogram are sought. The search uses the selection of adjacent vectors based on the recalculation of coordinates from the N-dimensional space to a one-dimensional space and vice versa and on the comparison with coordinates of neighboring vectors. Thereby the algorithm finds the cluster vertices in a single pass, and it does not require the creation of numerous local lists or vector adjacency graphs, which improves the efficiency of the clustering algorithm. Experimental results on automatic grouping of machine parts on the basis of drawings are discussed.

中文翻译：

---

基于工程图的一组机器零件聚类算法

摘要

对于工业物流任务，提出了一种基于工程图对一组给定的机器零件进行聚类的算法。为了加速聚类，对每个参数的值进行模糊处理，并寻找N维直方图的局部最大值。搜索根据N个坐标的重新计算使用相邻向量的选择维空间到一维空间，反之亦然，并与相邻向量的坐标进行比较。因此，该算法可在一次遍历中找到聚类顶点，并且不需要创建大量的本地列表或矢量邻接图，从而提高了聚类算法的效率。讨论了根据图纸对机器零件自动分组的实验结果。