文献类型：期刊文献

英文题名：Flow patterns classification of dense-phase pneumatic conveying of pulverized coal in the industrial vertical pipeline

作者：Jin, Yong[1];Lu, Haifeng[1];Guo, Xiaolei[1];Gong, Xin[1]

机构：[1]East China Univ Sci & Technol, Key Lab Coal Gasificat & Energy Chem Engn, Minist Educ, Shanghai Engn Res Ctr Coal Gasificat, Shanghai 200237, Peoples R China

年份：2019

卷号：30

期号：7

起止页码：1277

外文期刊名：ADVANCED POWDER TECHNOLOGY

收录：;EI(收录号：20191706828904);WOS:【SCI-EXPANDED(收录号:WOS:000468087700002)】；

基金：This work was supported by the National Key R&D Program of China (2018YFC0808503), the National Natural Science Foundation of China (51876066), the Shanghai Engineering Research Center of Coal Gasification (18DZ2283900) and the Fundamental Research Funds for the Central Universities.

语种：英文

外文关键词：Flow patterns; Pneumatic conveying; Classification model; Machine learning; Signal analysis

摘要：Dense-phase pneumatic conveying of pulverized coal experiments with superficial gas velocity ranged from 0.59 m s(-1) to 2.43 m s(-1) were carried out on the self-established 50 mm I.D. industrial pipe facility. The fluctuations of pressure, solid velocity, solid concentration and the average cross-section concentration were analyzed in terms of relative standard deviation (RSD), powder spectrum density analysis (PSD) and wavelet analysis. Further, a novel classification model of flow patterns for dense-phase pneumatic conveying of pulverized coal was proposed. The model was based on the relationships among superficial gas velocity, particle velocity and solid concentration. K-means method, std method, k-medoids and CART method were compared firstly and then the k-means method and CART method were introduced to classify the flow patterns. The bagged trees method and 5-fold cross-validation were used to train the classification model, whose accuracy was 95.2% and 91.9%. It meant that the new classification model was able to classify the flow patterns in the pneumatic conveying of pulverized coal. In the end, the change of flow patterns and transmission of flow patterns were explained by the wavelet analysis method from the aspects of different signal scales and power spectrum analysis. The new classification model will be of great help for the industry to have a better understanding on the classification and control of flow patterns. (C) 2019 The Society of Powder Technology Japan. Published by Elsevier B.V. and The Society of Powder Technology Japan. All rights reserved.