[1]杜锦程、吴福森、陈丙三.LS-SVM的核参数对概率筛筛分效率预测影响[J].福建工程学院学报,2021,19(01):12-18.[doi:10.3969/j.issn.1672-4348.2021.01.003]
　DU Jincheng,WU Fusen,CHEN Bingsan.Research on the influence of LS-SVM kernel parameters on the accurate prediction of probabilistic screening efficiency[J].Journal of FuJian University of Technology,2021,19(01):12-18.[doi:10.3969/j.issn.1672-4348.2021.01.003]
点击复制

LS-SVM的核参数对概率筛筛分效率预测影响()
分享到：

《福建工程学院学报》[ISSN:2097-3853/CN:35-1351/Z]

卷:
第19卷
期数:
2021年01期
页码:
12-18
栏目:
出版日期:
2021-02-25

文章信息/Info

Title:
Research on the influence of LS-SVM kernel parameters on the accurate prediction of probabilistic screening efficiency
作者:
杜锦程、吴福森、陈丙三
福建工程学院机械与汽车工程学院
Author(s):
DU Jincheng;  WU Fusen;  CHEN Bingsan
School of Mechanical and Automotive Engineering, Fujian University of Technology
关键词:
概率筛; LS-SVM分类算法; 核函数; 筛分效率预测建模
Keywords:
probability sieve;  LS-SVM classification algorithm;  kernel function;  screening efficiency prediction modeling
分类号:
TD452
DOI:
10.3969/j.issn.1672-4348.2021.01.003
文献标志码:
A
摘要:
以探索概率筛振动参数与筛分效率之间的关系，为概率筛结构的进一步改进提供指导意义为研究目的，将LS-SVM分类算法引入自同步概率筛筛分效率预测建模，探讨LS-SVM建模的可行性。基于各个不同的应用领域，可以构造不同的核函数，针对核函数需要优化特征参数的问题，应用网格搜索和交叉验证算法，对核参数的选择进行优化。通过研究得出用多项式(Poly)核函数建模对预测样本的最高预测识别率达到96.7%，采用RBF核函数建模对预测样本达到了零错分率，表明将LS-SVM算法引入概率筛筛分效率预测建模是可行的。
Abstract:
In order to explore the relationship between the vibration parameters of probabilistic screening and the screening efficiency, and to provide guidance for the further improvement of the probabilistic screening structure, LS-SVM classification algorithm was introduced into the self-synchronous probabilistic screening efficiency prediction modeling, and the feasibility of LS-SVM modeling was discussed. Based on different application fields, different kernel functions can be constructed. To solve the problem that kernel functions need to optimize characteristic parameters, grid search and cross validation algorithms were applied to optimize the selection of kernel parameters. Results show that the highest predictive recognition rate of the predicted samples by using Poly kernel function modeling is 96.7%, and the zero error rate of the predicted samples by using RBF kernel function modeling indicates that it is feasible to introduce LS-SVM algorithm into probabilistic screening efficiency prediction modeling.

参考文献/References:

[1] 焦红光, 赵跃民. 用颗粒离散元法模拟筛分过程[J]. 中国矿业大学学报, 2007, 36（2）: 232-236.[2] 康娅娟, 姚尧, 黄毅, 等. 多层概率筛筛分工艺性能评价方法研究[J]. 机械设计, 2017, 34（1）: 66-70.[3] DAVOODI A, BENGTSSON M, HULTHN E, et al. Effects of screen decks’ aperture shapes and materials on screening efficiency[J]. Minerals Engineering, 2019, 139: 105699.[4] SAFRANYIK F, CSIZMADIA B M, HEGEDUS A, et al. Optimal oscillation parameters of vibrating screens[J]. Journal of Mechanical Science and Technology, 2019, 33（5）: 2011-2017.[5] 郑桂霞, 黄宜坚. 自同步概率筛效率研究[J]. 机械科学与技术, 2010, 29（7）: 944-947, 952.[6] VAPNIK V N. The nature of statistical learning theory[M]. New York: Springer, 1995.[7] CHERKASSKY V, Mulier F. Learning from data: concepts, theory and methods[M]. New York:John Viley & Sons, 1997.[8] 汤琴, 黄宜坚. 采用AR模型双谱估计的概率筛筛分效率[J]. 华侨大学学报（自然科学版）, 2011, 32（3）: 253-257.[9] VAPNIK V. Estimation of regression parameters[M]∥Estimation of Dependences Based on Empirical Data. New York: Springer, 2006: 109-138.[10] SUYKENS J, VANDEWALLE J. Least squares support vector machine classifiers[J]. Neural Processing Letters, 1999, 9（3）: 293-300.[11] ZHANG J. The sample breakdown points of tests[J]. Journal of Statistical Planning and Inference, 1996, 52（2）: 161-181.

更新日期/Last Update: 2021-02-25