基于聚类分析的育种小麦机械裂纹硬度算法

Algorithm of breeding wheat mechanical crack hardness based on clustering analysis

为进一步寻求适宜作为育种小麦机械裂纹硬度值聚类分析的优势算法,通过借助于虚拟弹性模量计算法,测定产生机械裂纹的育种小麦籽粒不同组分硬度值,并结合基于邻域改进K-medoids算法与传统的PAM算法、快速K-medoids算法分别对育种小麦籽粒(表皮、内部)机械裂纹硬度数据集进行聚类分析比较。实验结果表明:改进K-medoids算法的时间性能明显优于PAM算法和快速K-medoids算法,在育种小麦籽粒表皮、内部机械裂纹硬度数据分析时间上分别减少2.585s、0.103s和0.603s、0.551s,较其余两种算法的聚类误差平方和小、聚类准确率高(90%以上),该算法可为育种小麦籽粒信息的准确、快速整理提供理论依据。

For seeking superior algorithm of clustering analysis for breeding wheat mechanical crack hardness, improved algorithm of K-medoids was selected, which can provide theoretical basis for information statistics of breeding wheat grain. Comparison of clustering analysis of breeding wheat grain （epidermis and inside） mechanical crack hardness datasets among improved algorithm of K-medoids, fast K-medoids algorithm and the traditional PAM algorithms were conducted, by using virtual elastic modulus calculation method to testing the mechanical cracks hardness value. The experimental results show that time performance of improved K medoids algorithm is better than the other two, the analyzing time for hardness data of grain epidermis and inside reduced 2. 585s, 0. 103s and 0. 603s, 0. 551s respectively. Quadratic sum of clustering error of improved K medoids algorithm is smaller and the clustering accura cy is higher, which can achieve above 90%.