采用多变量RBF神经网络的非线性内部迭代预测控制

doi:10.3969/j.issn.0255-8297.2018.04.013

应用科学学报 ›› 2018, Vol. 36 ›› Issue (4): 698-710.doi: 10.3969/j.issn.0255-8297.2018.04.013

采用多变量RBF神经网络的非线性内部迭代预测控制

姜雪莹¹, 苏成利¹, 施惠元^1,2, 李平^1,2, 刘思雨¹

1. 辽宁石油化工大学信息与控制工程学院, 辽宁抚顺 113001;
2. 西北工业大学自动化学院, 西安 710072

收稿日期:2017-01-19 修回日期:2017-05-14 出版日期:2018-07-31 发布日期:2018-07-31
通信作者: 苏成利,教授,研究方向:模型预测控制、先进过程控制、智能控制与优化技术等,E-mail:sclwind@sina.com E-mail:sclwind@sina.com
基金资助:
国家自然科学基金（No.61673199）；辽宁省高等学校优秀科技人才支持计划项目基金（No.LJQ2015061，No.LR2015034）；受工业控制技术国家重点实验室开放课题基金（No.ICT1800400）资助

Nonlinear Internal Iterative Predictive Control Using Multi-RBF Neural Network

JIANG Xue-ying¹, SU Cheng-li¹, SHI Hui-yuan^1,2, LI Ping^1,2, LIU Si-yu¹

1. School of Information and Control Engineering, Liaoning Shihua University, Fushun 113001, Liaoning Province, China;
2. School of Automation, Northwestern Polytechnical University, Xi'an 710072, China

Received:2017-01-19 Revised:2017-05-14 Online:2018-07-31 Published:2018-07-31

摘要/Abstract

摘要： 针对工业过程中被控对象往往具有复杂、强非线性、多变量的特性，提出一种基于多变量径向基函数（radial base function，RBF）神经网络的非线性内部迭代预测控制方法.采用多个RBF神经网络在线逼近多输入多输出的非线性系统，得到一个近似模型作为预测模型.同时为了减少求解系统控制律的计算量，将每个输出预测值沿着输入轨迹展开，从而把求解复杂非线性优化问题转化为求解简单的二次规划问题，解决了在线实时递推控制律时求解非线性微分方程的困难.最后通过t次内部迭代直至满足迭代条件，得到了最优的控制律.pH中和过程的仿真结果表明了该算法是有效而可行的.

关键词: 径向基函数神经网络, 内部迭代, 多变量, 预测控制

Abstract: For the complexity, strong nonlinearity and multi-variability in industrial processes, a nonlinear iterative predictive control algorithm based on radial base function (RBF) neural network is proposed. This algorithm employs multi-RBF neural network to approximate the nonlinear system, which can obtain an approximate model for the predictive model. Meanwhile in order to avoid increasing some computational burden, each predictive output along the future trajectory is expanded. Therefore the problem of nonlinear optimization is transformed into that of solving the easy quadratic programming,accordingly, thus the difficulty of solving the nonlinear differential equation in real-time online can be overcome, when deriving the predictive control law. Finally, the optimized control law can be obtained until the internal conditions are satisfied through internal iteration of t times. The simulation results for a pH process show that the proposed method is effective and feasible.

Key words: radial base function (RBF) neural network, predictive control, internal iterative, multi-variable

中图分类号:

P751.1

姜雪莹, 苏成利, 施惠元, 李平, 刘思雨. 采用多变量RBF神经网络的非线性内部迭代预测控制[J]. 应用科学学报, 2018, 36(4): 698-710.

JIANG Xue-ying, SU Cheng-li, SHI Hui-yuan, LI Ping, LIU Si-yu. Nonlinear Internal Iterative Predictive Control Using Multi-RBF Neural Network[J]. Journal of Applied Sciences, 2018, 36(4): 698-710.

参考文献

[1] Wang L P, Young P C. An improved structure for model predictive control using non-minimal state space realization[J]. Journal of Process Control, 2006, 16(4):355-371.
[2] Yang B, Xu Z H, Yang Y, Gao F R. Application of two-dimensional predictive functional control in injection molding[J]. Relations Internationales, 2015, 133(1):41-52.
[3] 苏成利,施惠元,李平,王倩,许娜. 增量式多变量预测函数控制及其在气体分馏装置中的应用[J]. 南京理工大学学报,2015, 39(5):625-631. Su C L, Shi H Y, Li P, Wang Q, Xu N. Incremental mutivariable predictive control and its application in a gas fractionation unit[J]. Journal of Nanjing University of Science and Technology, 2015, 39(5):625-631. (in Chinese)
[4] Su C L, Shi H Y, Li P, Cao J T. Advanced control in a delayed coking furnace[J]. Measurement and Control, 2015, 48(2):54-59.
[5] 谢亚军,丁宝苍,陈桥. 状态空间模型的双层结构预测控制算法[J]. 控制理论与应用,2017, 34(1):69-76. Xie Y J, Ding B C, Chen Q. Double-layered model predictive control of state-space model[J]. Control Theory & Applications, 2017, 34(1):69-76. (in Chinese)
[6] 黄骅,何德峰,俞立. 基于多面体描述系统的鲁棒非线性预测控制[J]. 自动化学报,2012, 38(12):1906-1912. Huang H, He D F, Yu L. Robust nonlinear predictive control based on polytopic description systems[J]. Acta Automatica Sinica, 2012, 38(12):1906-1912. (in Chinese)
[7] 陈进东,潘丰. 基于在线支持向量回归的非线性模型预测控制方法[J]. 控制与决策,2014(3):460-464. Chen J D, Pan F. Online support vector regression-based nonlinear model predictive control[J]. Control and Decision, 2014(3):460-464. (in Chinese)
[8] He D, Wang L, Yu L. Multi-objective nonlinear predictive control of process systems:a dual-mode tracking control approach[J]. Journal of Process Control, 2015, 25:142-151.
[9] Shi H Y, Su C L, Cao J T, Li P, Liang J P, Zhong G C. Nonlinear adaptive predictive functional control based on the Takagi-Sugeno model for average cracking outlet temperature of the ethylene cracking furnace[J]. Industrial Engineering Chemistry Research, 2015, 54(6):1849-1860.
[10] Kang J, Meng W. System identification based on improved BP neural networks[J]. Journal of Computational Information Systems, 2012, 8(5):2099-2106.
[11] 李晓华,李军. 基于ESN网络的连续搅拌反应釜(CSTR)辨识[J]. 信息与控制,2014, 43(2):223-228. Li X H, Li J. Identification of continuous stirred tank reactor (CSTR) based on echo state network (ESN)[J]. Information and Control, 2014, 43(2):223-228. (in Chinese)
[12] Nidhil K J, Sreeraj S, Vijay B, Bagyaveereswaran V. System identification using artificial neural network[C]//2015 International Conference on Circuit, Power and Computing Technologies. IEEE, 2015:1-4.
[13] Tatjewski P. Advanced control of industrial processes[M]. London:Springer, 2007.
[14] Jing N, Ren X, Cong S, Shang C, Guo Y. Adaptive neural network predictive control for nonlinear pure feedback systems with input delay[J]. Journal of Process Control, 2012, 22(1):194-206.
[15] Nikdel N, Nikdel P, Badamchizadeh M A, Hassaniadeh I. Using neural network model predictive control for controlling shape memory alloy-based manipulator[J]. IEEE Transactions on Industrial Electronics, 2014, 61(3):1394-1401.
[16] Lin Y C, Chen D D, Chen M S, Chen X M, Li J. A precise BP neural network-based online model predictive control strategy for die forging hydraulic press machine[J]. Neural Computing & Applications, 2016:1-12.
[17] 苏成利,吴云,刘晓琴. 一种基于PSO的自适应神经网络预测控制[J]. 控制工程,2009, 16(4):454-457. Su C L, Wu Y, Liu X Q. A PSO-based adaptive neural network model predictive control[J]. Control Engineering, 2009, 16(4):454-457. (in Chinese)
[18] 李会军,肖兵. 一种无约束多步递归神经网络预测控制器[J]. 控制理论与应用,2012, 29(5):642-648. Li H J, Xiao B. Multistep recurrent neural network model predictive controller without constraints[J]. Control Theory & Applications, 2012, 29(5):642-648. (in Chinese)
[19] Łwryńzuk m. Accuracy and computational efficiency of suboptimal nonlinear predictive control based on neural models[J]. Applied Soft Computing, 2011, 11(2):2202-2215.
[20] Gomez J C, Jutan A, Baeyens E. Wiener model identification and predictive control of a pH neutralisation process[J]. IEE Proceedings-Control Theory and Applications, 2004, 151(3):329-338.

采用多变量RBF神经网络的非线性内部迭代预测控制

Nonlinear Internal Iterative Predictive Control Using Multi-RBF Neural Network

PDF

可视化

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 12

编辑推荐

Metrics

本文评价

[1]	魏巍, 全海燕. 基于单形进化的径向基网络训练算法[J]. 应用科学学报, 2019, 37(4): 459-468.
[2]	王德明, 隋修武, 张杨, 万凯新, 石峰. 基于DMA与改进GPC算法的超细缝合线直径控制技术[J]. 应用科学学报, 2019, 37(3): 349-358.
[3]	姜雪莹, 施惠元, 苏成利, 李平. 基于TSA的非线性神经网络预测控制[J]. 应用科学学报, 2018, 36(5): 870-878.
[4]	王建宏1，许莺1，熊朝华1，徐波2. 故障条件下子空间预测控制的对偶分解[J]. 应用科学学报, 2014, 32(6): 652-660.
[5]	田中大，高宪文，李琨. 网络控制系统的自适应预测控制[J]. 应用科学学报, 2013, 31(3): 303-308.
[6]	焉晓贞，谢红，王桐. 无线传感器网络的不确定传感数据预测[J]. 应用科学学报, 2012, 30(6): 566-572.
[7]	王建宏，王道波. 子空间预测控制中的椭球优化及其应用[J]. 应用科学学报, 2010, 28(4): 424-429.
[8]	邹涛, 闫志恒, 李少远. 基于PC机平台的多变量预测控制软件开发[J]. 应用科学学报, 2003, 21(3): 296-301.
[9]	杨涛, 王学雷, 邵惠鹤. 基于在线训练RBF神经网络的混沌系统辨识研究[J]. 应用科学学报, 2002, 20(2): 160-164.
[10]	李一波, 黄小原. 基于RBFNN和GA的重叠峰分辨新技术[J]. 应用科学学报, 2002, 20(1): 99-103.
[11]	陈增强, 袁著祉. PI型有约束后移限位预测控制器[J]. 应用科学学报, 1998, 16(2): 229-233.
[12]	陈增强, 袁著祉. PI型扩展预测自适应控制器及其在造纸过程中的应用研究[J]. 应用科学学报, 1996, 14(4): 447-452.