• 主题
高级搜索  >
历史搜索 清空历史
首页> 外文期刊>Arabian Journal for Science and Engineering. Section A, Sciences >SVM-Based Predictive Modelling ofWet Pelletization Using Experimental and GA-Based Synthetic Data
【24h】

SVM-Based Predictive Modelling ofWet Pelletization Using Experimental and GA-Based Synthetic Data

机译：基于SVM的预测建模WET使用实验和基于GA合成数据的造粒

获取原文
获取原文并翻译 | 示例
           
页面导航
  • 摘要
  • 著录项
  • 相似文献
  • 相关主题

摘要

A soft computing-based approach to capture the influence of process parameters on wet pelletization is presented. Wet pelletization is highly susceptible to even the minute change in operating conditions that results in substantial changes in product characteristics. This sensitive nature inspired research towards the development of new models for forecasting pellet attributes and increasing process understanding. The present modelling approach is based on support vector machine (SVM), a soft computing technique, to predict wet pellet characteristics and to analyse the effect of individual variable parameters. Experiments are carried out on a laboratory scale disc pelletizer with changing parameter values. However, experimental data were insufficient to train SVM and afterwards to test its prediction accuracy. To accomplish this task, genetic algorithm is employed to generate synthetic data and its fitness functions are derived using multiple regression. SVM is trained and validated on synthetic data while original data are used for testing. The significance of process parameters on pellet characteristics is also examined by deliberately removing them one by one from the model and recording the corresponding variation in prediction accuracy. The study employing SVM for the first time establishes the predictive ability of SVM with precision in wet pelletization.
机译：介绍了捕获基于软计算的方法来捕获过程参数对湿造粒的影响。甚至均衡的湿法化甚至可能导致产品特性的显着变化的操作条件下的微小变化。这种敏感的性质启发了对预测颗粒属性的新模型的开发和增加过程理解的研究。本发明的建模方法基于支持向量机（SVM），软计算技术，预测湿颗粒特性并分析各个变量参数的效果。实验在实验室刻度盘造粒机上进行，具有改变参数值。然而，实验数据不足以训练SVM和之后的测试以测试其预测准确性。为了完成此任务，使用遗传算法来生成合成数据，并且使用多元回归来导出其健身功能。 SVM在合成数据上培训并验证，而原始数据用于测试。还通过故意从模型中逐一移除它们并记录预测精度的相应变化来检查过程参数对颗粒特性的重要性。第一次使用SVM的研究建立了SVM在湿造粒中精确的预测能力。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱：kefu@zhangqiaokeyan.com

京公网安备：11010802029741号 ICP备案号：京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有

  • 客服微信

  • 服务号