磨粒特征参数的筛选及权分析

基于概率分布特性及集合论 ,提出区分度概念并建立一种用于对金属磨粒的特征参数进行分析筛选的方法。该方法根据特征参数的数据结构特征 ,筛选出具有较强分类特性的特征参数 ,而将分类特性差的或是派生的特征参数能准确地分离出来。本文同时采用集合域分析方法 ,对特征参数进行了权分析 ,从而大大提高了特征参数的分类特性。文中对金属磨粒分类中常用的几种几何特征参数 (圆度、长 /短轴比、线性度、凹度及综合参数 )进行了计算分析 ,结果显示圆度的区分度最大 ,其他依次为长 /短轴比、综合参数、凹度和线性度。而这一结果与作者根据实际经验来对磨粒进行分类所采用的分类决策的秩序是相一致的 ,从而说明该方法的有效性及实用性

Screen and Weight Analysis of Wear Particle Characteristic Parameters

Based on the probability and set theories, a new conception distinguishable degree is built. A new approach is also given to analyze screen and wear particle characteristic parameters. In this approach, the better classification parameters can be selected, and the poorer classification parameters can be screened from the set of characteristic parameters. Using the set field, the weight analysis of the parameters is carried out, and the classification of the parameters is improved. Meanwhile some geometrical parameters used in the classification of the wear particles are calculated and analyzed. Result shows that the round degree is the best classification, and then the ratio of long to short pivot, the synthetical parameter, the concave degree and the line degree in their proper order. Result corresponds with the classification order used in the real experience. Therefore, it can conclude that the approach is effective and feasible.