CIESC Journal

Previous Articles     Next Articles

Multi-delays identification for alumina evaporation process based on time-correlation analysis

WANG Feng, WANG Xiaoli, XIE Yongfang, XIE Sen, YANG Chunhua   

  1. School of Information Science and Engineering, Central South University, Changsha 410083, Hunan, China
  • Received:2016-11-14 Revised:2016-11-18 Online:2017-03-05 Published:2017-03-05
  • Contact: 10.11949/j.issn.0438-1157.20161604
  • Supported by:

    supported by the National High Technology Research and Development Program of China (2014AA041803).

基于时效关联分析的氧化铝蒸发过程多重时滞辨识

王峰, 王晓丽, 谢永芳, 谢森, 阳春华   

  1. 中南大学信息科学与工程学院, 湖南 长沙 410083
  • 通讯作者: 谢永芳,yfxie@csu.edu.cn
  • 基金资助:

    国家高技术研究发展计划项目(2014AA041803)。

Abstract:

The multi-delays of the whole equipment unit in the process industry are indicated in the form of a sequence of integers. A restructured data matrix is generated based on the sequence and the original data. The time-correlation analysis is defined to describe the correlation between the columns in the data matrix and H norm is used to quantify the correlation of the matrix so that the multi-delays identification problem can be converted to calculating the biggest H norm. At last, the discrete state transition algorithm is adopted to quickly search the biggest H norm and the multi-delays are identified. The proposed method is used for multi-delays identification of alumina evaporation process and the result of multi-delays identification is applied to the data preprocessing of a prediction model. The accuracy of the prediction model is improved by 34.4% and the proposed identification method is effective.

Key words: alumina, evaporation, time-correlation analysis, multi-delays identification, discrete state transition algorithm, H norm, prediction

摘要:

将流程工业中物料在每个单元的滞留时间以整数序列的形式表示,根据时滞序列将原始数据重新匹配生成新的数据矩阵。定义时效关联分析来描述数据矩阵中各个时间序列之间的相关性并使用H范数进行量化,将多重时滞辨识问题转化为求取最大H范数的问题。最后采用离散状态转移算法进行快速求解,得到最优时基序列即为辨识得到的多重时滞。将所提方法用于氧化铝蒸发过程的多重时滞辨识,将辨识出的多重时滞用于预测模型数据的预处理,预测模型的准确性提高了34.4%,验证了辨识方法的有效性。

关键词: 氧化铝, 蒸发, 时效关联分析, 多重时滞辨识, 离散状态转移算法, H范数, 预测

CLC Number: