高级检索

  • ISSN 1006-3080
  • CN 31-1691/TQ

基于新型自适应采样算法的催化重整过程代理模型

张剑超 杜文莉 覃水

张剑超, 杜文莉, 覃水. 基于新型自适应采样算法的催化重整过程代理模型[J]. 华东理工大学学报(自然科学版), 2019, 45(6): 928-937. doi: 10.14135/j.cnki.1006-3080.20180915001
引用本文: 张剑超, 杜文莉, 覃水. 基于新型自适应采样算法的催化重整过程代理模型[J]. 华东理工大学学报(自然科学版), 2019, 45(6): 928-937. doi: 10.14135/j.cnki.1006-3080.20180915001
ZHANG Jianchao, DU Wenli, QIN Shui. Surrogate Model of Catalytic Reforming Process Based on a New Adaptive Sampling Algorithm[J]. Journal of East China University of Science and Technology, 2019, 45(6): 928-937. doi: 10.14135/j.cnki.1006-3080.20180915001
Citation: ZHANG Jianchao, DU Wenli, QIN Shui. Surrogate Model of Catalytic Reforming Process Based on a New Adaptive Sampling Algorithm[J]. Journal of East China University of Science and Technology, 2019, 45(6): 928-937. doi: 10.14135/j.cnki.1006-3080.20180915001

基于新型自适应采样算法的催化重整过程代理模型

doi: 10.14135/j.cnki.1006-3080.20180915001
基金项目: 国家自然科学基金重大项目(61590923);国家自然科学基金杰出青年项目(61725301);国家自然科学基金青年项目(21506050);中央高校基本科研业务费
详细信息
    作者简介:

    张剑超(1994-),男,山东青岛人,硕士生,研究方向为化工过程的代理模型。E-mail:532978533@qq.com

    通讯作者:

    杜文莉,E-mail:wldu@ecust.edu.cn

  • 中图分类号: TE624

Surrogate Model of Catalytic Reforming Process Based on a New Adaptive Sampling Algorithm

  • 摘要: 催化重整装置是石油加工的重要装置之一,直接使用机理模型对其进行优化耗时较长。代理模型方法能够有效地对机理模型进行近似,而采样方法对代理模型的精度有很大影响。提出了一种新的自适应采样算法−基于最近邻和马氏距离的自适应采样算法。该算法从采样方法的全局搜索能力与局部搜索能力出发,通过求解优化问题在关键信息区域中获取样本点。采用7个测试函数进行测试,结果表明该算法能够选取对代理模型精度影响较大的采样点,从而有效地提升代理模型的精度。针对催化重整过程的关键质量指标建立了相应的代理模型,结果表明该算法能够很好地处理实际工程中的问题。

     

  • 图  1  关键信息区域图例

    Figure  1.  An example of informative region

    图  2  ASA-NNMD流程图

    Figure  2.  Flowchart of the ASA-NNMD

    图  3  PK函数的真实函数图

    Figure  3.  Real image of PK function

    图  4  ASA-NNMD 的采样点分布图

    Figure  4.  Sampling point distribution maps of ASA-NNMD

    图  5  EIGF的采样点分布图

    Figure  5.  Sampling point distribution maps of EIGF

    图  6  MASA的采样点分布图

    Figure  6.  Sampling point distribution maps of MASA​​​​​

    图  7  连续催化重整装置流程图

    Figure  7.  Flowchart of continuous catalytic reforming

    图  8  芳烃收率(a)、重整液收率(b)、纯氢收率(c)的RMSE随迭代次数的变化曲线

    Figure  8.  Variation curves of RMSE of aromatics yield (a), liquid yield (b), hydrogen yield (c) with numbers of iteration

    图  9  芳烃收率(a)、重整液收率(b)、纯氢收率(c)的MAE随迭代次数的变化曲线

    Figure  9.  Variation curves of MAE of aromatics yield (a), liquid yield (b), hydrogen yield (c) with numbers of iteration

    表  1  7个测试函数的采样方式

    Table  1.   Sampling approaches of seven test functions

    Test functionnInitial numberGenerationTest number
    SH220801 000
    PK220801 000
    SC220801 000
    GP220801 000
    DP4401601 500
    Shekel54401601 500
    Hart66602402 000
    下载: 导出CSV

    表  2  7个测试函数的平均RMSE

    Table  2.   Average RMSE of seven test functions

    Test functionRMSE
    ASA-NNMDEIGFMASA
    SH0.015 759 30.017 662 20.016 153 4
    PK0.137 228 70.143 066 20.379 633 2
    SC0.047 608 90.077 321 30.054 711 8
    GP5 146.485 536 6 119.994 726 5 958.324 394 1
    DP5 708.543 027 172.135 386 8 388.879 769
    Shekel50.082 294 70.085 797 50.091 177 2
    Hart60.211 107 10.237 356 70.269 509 9
    下载: 导出CSV

    表  3  7个测试函数的平均MAE

    Table  3.   Average MAE of seven test functions

    Test functionMAE
    ASA-NNMDEIGFMASA
    SH0.058 616 0.123 294 0.046 424
    PK0.074 098 0.070 576 0.232 003
    SC0.000 245 0.000 703 0.000 181
    GP3 340.0073 438.9014 540.451
    DP3 201.215 576.815 450.233
    Shekel50.041 845 0.071 110.056 102
    Hart60.116 427 0.144 379 0.130 05
    下载: 导出CSV

    表  4  7个测试函数的平均计算时间

    Table  4.   Average running time of seven test functions

    Test functiont/s
    ASA-NNMDEIGFMASA
    SH206.965.6152.2
    PK201.165.5156.8
    SC201.865.8158.9
    GP199.166.2155.9
    DP917.5208.9320.3
    Shekel5923.8219.5307.1
    Hart63 025.3495.9517.9
    下载: 导出CSV

    表  5  设计变量取值范围

    Table  5.   Upper bound and lower bound of design variables

    TR1/℃TR2/℃TR3/℃TR4/℃RH/OF/(t·h−1)
    515~525515~525515~525515~5252.5~3120~170
    下载: 导出CSV

    表  6  3个主要质量指标的RMSE

    Table  6.   RMSE of three quality indexs

    Sampling approachRMSE
    Aromatics yieldLiquid yieldHydrogen yield
    ASA-NNMD1.14×10−41.56×10−43.1×10−5
    EIGF1.57×10−41.58×10−43.4×10−5
    MASA1.94×10−41.78×10−43.3×10−4
    下载: 导出CSV

    表  7  3个主要质量指标的MAE

    Table  7.   MAE of three quality indexs

    Sampling approachMAE
    Aromatics yieldLiquid yieldHydrogen yield
    ASA-NNMD8.1×10−58.2×10−52.2×10−5
    EIGF1.36×10−41.05×10−42.5×10−5
    MASA1.17×10−41.35×10−52.4×10−5
    下载: 导出CSV
  • [1] FORRESTER A I J, KEANE A J. Recent advances in surrogate-based optimization[J]. Progress in Aerospace Sciences, 2009, 45(1): 50-79.
    [2] NEMETH M A, BORROR C M. Response Surface Methodology: Process and Product Optimization Using Designed Experiments [M]. New York, USA: John Wiley & Sons Inc, 2003.
    [3] QUEIPO N V, HAFTKA R T, WEI S, et al. Surrogate-based analysis and optimization[J]. Progress in Aerospace Sciences, 2005, 41(1): 1-28. doi: 10.1016/j.paerosci.2005.02.001
    [4] CRESSIE N. Spatial prediction and ordinary Kriging[J]. Mathematical Geology, 1989, 21(4): 493-494. doi: 10.1007/BF00897332
    [5] FORRESTER D A I J, SÓBESTER D A, Keane A J. Engineering Design via Surrogate Modelling: A Practical Guide [M]. New York, USA: John Wiley & Sons Inc, 2008.
    [6] CARL EDWARD R. Gaussian Processes for Machine Learning [M]. USA: MIT Press, 2006: 69-106.
    [7] DYN N, LEVIN D, RIPPA S. Numerical procedures for surface fitting of scattered data by radial functions[J]. SIAM Journal on Scientific and Statistical Computing, 1986, 7(2): 639-659. doi: 10.1137/0907043
    [8] HUSSAIN M F, BARTON R R, JOSHI S B. Metamodeling: Radial basis functions, versus polynomials[J]. European Journal of Operational Research, 2007, 138(1): 142-154.
    [9] CLARKE S M, GRIEBSCH J H, SIMPSON T W. Analysis of support vector regression for approximation of complex engineering analyses[C]// ASME 2003 International Design Engineering Technical Conferences and Computers and Information in Engineering Conference. USA: American Society of Mechanical Engineers, 2005: 1077-1087.
    [10] BASHEER I A, HAJMEER M. Artificial neural networks: Fundamentals, computing, design, and application[J]. Journal of Microbiological Methods, 2000, 43(1): 3-31. doi: 10.1016/S0167-7012(00)00201-3
    [11] LIU H, ONG Y S, CAI J. A survey of adaptive sampling for global metamodeling in support of simulation-based complex engineering design[J]. Structural & Multidisciplinary Optimization, 2017, 57(1): 393-416.
    [12] CROMBECQ K, GORISSEN D, DESCHRIJVER D, et al. A novel hybrid sequential design strategy for global surrogate modeling of computer experiments[J]. SIAM Journal on Scientific Computing, 2011, 33(4): 1948-1974. doi: 10.1137/090761811
    [13] ZHOU B D, YAO H L, SHI M H, et al. An new immune genetic algorithm based on uniform design sampling[J]. Knowledge & Information Systems, 2012, 31(2): 389-403.
    [14] ZHANG Y S, LI W G, MAO S S, et al. Orthogonal arrays obtained by generalized difference matrices with g levels[J]. Science China Mathematics, 2011, 54(1): 133-143. doi: 10.1007/s11425-010-4144-y
    [15] MCKAY M D, BECKMAN R J, CONOVER W J. A comparison of three methods for selecting values of input variables in the analysis of output from a computer code[J]. Technometrics, 2000, 42(1): 55-61. doi: 10.1080/00401706.2000.10485979
    [16] GARUD S S, KARIMI I A, KRAFT M. Design of computer experiments: A review[J]. Computers & Chemical Engineering, 2017, 106: 71-95.
    [17] WANG G G, SHAN S. Review of metamodeling techniques in support of engineering design optimization[C]// ASME 2006 International Design Engineering Technical Conferences and Computers and Information in Engineering Conference. USA: American Society of Mechanical Engineers, 2007: 415-426.
    [18] ANGLUIN D. Queries and concept learning[J]. Machine Learning, 1987, 2(4): 319-342.
    [19] SHEWRY M C, WYNN H P. Maximum entropy sampling[J]. Journal of Applied Statistics, 2015, 14(2): 165-170.
    [20] JIN R, CHEN W, SUDJIANTO A. On sequential sampling for global metamodeling in engineering design[C]//ASME 2002 International Design Engineering Technical Conferences and Computers and Information in Engineering Conference. Montreal, Canada: ASME, 2002: 539-548
    [21] CROMBECQ K, DE TOMMASI L, GORISSEN D, et al. A novel sequential design strategy for global surrogate modeling[C]//Winter Simulation Conference. Austin: ACM, 2009: 734-742.
    [22] YANG Q, XUE D. A weighted sequential sampling method considering influences of sample qualities in input and output parameter spaces for global optimization[J]. Journal of Optimization Theory & Applications, 2015, 164(2): 644-665.
    [23] SACKS J, WELCH W J, MITCHELL T J, et al. Design and analysis of computer experiments[J]. Statistical Science, 1989, 94(4): 307-309.
    [24] LOPHAVEN S N, NIELSEN H B, SONDERGAARD J. DACE—-A Matlab Kriging toolbox, version 2.0[R]. Denmark: Technical University of Denmark, 2002.
    [25] LIU H, CAI J, ONG Y S. An adaptive sampling approach for Kriging metamodeling by maximizing expected prediction error[J]. Computers & Chemical Engineering, 2017, 106: 171-182.
    [26] 杨绪兵, 王一雄, 陈斌. 马氏度量学习中的几个关键问题研究及几何解释[J]. 南京大学学报(自然科学版), 2013, 49(2): 133-141.
    [27] MAHALANOBIS P C. On the generalized distance in statistics[J]. Proceedings of the National Institute of Sciences, 1936, 2(1): 49-55.
    [28] 阴盼强, 路东明, 袁渊. 基于马氏距离的改进非局部均值图像去噪算法[J]. 计算机辅助设计与图形学学报, 2016, 28(3): 404-410. doi: 10.3969/j.issn.1003-9775.2016.03.004
    [29] EASON J, CREMASCHI S. Adaptive sequential sampling for surrogate model generation with artificial neural networks[J]. Computers & Chemical Engineering, 2014, 68: 220-232.
    [30] CHEN Q L. Sequential adaptive designs in computer experiments for response surface model fit [D]. USA: Ohio State University, 2008.
    [31] 罗雷涛. 连续催化重整过程建模方法研究及应用[D]. 上海: 华东理工大学, 2015.
    [32] 周红军. 芳烃型连续重整集总反应动力学模型研究[D]. 上海: 华东理工大学, 2011.
    [33] WEI M, YANG M, QIAN F, et al. Integrated dual-production mode modeling and multiobjective optimization of an industrial continuous catalytic naphtha reforming process[J]. Industrial & Engineering Chemistry Research, 2016, 55(19): 5714-5725.
  • 加载中
图(9) / 表(7)
计量
  • 文章访问数:  10385
  • HTML全文浏览量:  2749
  • PDF下载量:  33
  • 被引次数: 0
出版历程
  • 收稿日期:  2018-09-25
  • 网络出版日期:  2019-06-06
  • 刊出日期:  2019-12-01

目录

    /

    返回文章
    返回