基于ADRC的随机库存控制 本期目录 >>
Title: Stochastic Inventory Control Based on ADRC
作者 任庆忠;张 荣;邹莉娜
Author(s): REN Qing-zhong; ZHANG Rong; ZOU Li-na
摘要: 自抗扰控制(ADRC)是一种独立于模型的控制思想,尽管人们过去主要关注的是ADRC在工业控制领域中的应用,但其设计思路也比较自然地反映了决策者在动态优化问题中的理性分析过程。本文基于随机动态优化理论和ADRC两种不同方法,对比研究经典的随机生产库存模型。研究表明:当需求函数为常数时,ADRC给出的生产策略有近45%的概率比随机动态优化理论求解的生产策略带来的总成本更低;当需求函数含未知的季节性周期变动时,此时再利用随机最优控制理论无法求得最优生产策略的解析解,而ADRC仍然能够消除不确定环境带来的冲击,通过实时反馈获得满意的生产策略。因此,融入经济与管理科学问题的相关特点,合理地利用ADRC可能大大简化问题的分析难度,为随机动态优化问题提供一种新的分析思路。
Abstract: Many problems in economics and management science can be formulated as optimal control problems. Generally, we could use the approach of Pontryagin maximum principle to analyze them. This approach often involves in solving a two-point-boundary-value problem(TPBVP) which seldom admits closed-form expression for the optimal solution. In addiction, its analysis would become much more difficult if the uncertainty is introduced. Active disturbances rejection controller (ADRC) is a model-independent controller. It has good robustness and adaptability for the controlled systems with nonlinearity, large time delay, and high uncertainty. We notice that, although the previous work has focused mainly on its application in industrial engineering, the basic idea of ADRC also reflects very naturally the rational behavior of a decision maker in the face of dynamic optimization problems in economics and management science. This paper investigates a classical stochastic inventory control model under the framework of ADRC. First, we give a brief introduction of ADRC and the traditional approach used in the analysis of inventory control problems. Then we demonstrate how an ADRC can be used to analyze a specific inventory control problem. In this model, ADRC is composed of three parts, i.e., the tracking differentiator, extended state observer, and nonlinear state feedback. Given a target inventory level, the tracking differentiator is used as arrangements for the transition process, the extended state observer is used to obtain the actual output of the controlled system and to estimate its on-line derivatives of different orders, and the nonlinear state feedback control is constructed by the information obtained from the tracking differentiator and the extended state observer. Two typical cases, i.e., constant demand and demand with seasonal change are discussed in detail. The optimal production rates and the corresponding optimal inventory trajectories are obtained by ADRC and they are compared to those obtained by the traditional methods. Research shows that, if the demand function is a constant which is assumed to be known in advance, the total cost of production and inventory by the approach of ADRC would be lower with a probability of about 45% than the one obtained from a standard method of stochastic optimal control. This indicates that ADRC is highly effective in dealing with the controlled systems whose structures are completely known. On the other hand, if the demand function allows for an unknown seasonal element, the traditional method of dynamic optimization will no longer be valid for this case, but the approach of ADRC can still work very well to obtain satisfactory results. From a viewpoint of application, ADRC should be a powerful tool for solving a large class of control problems with or without uncertainties.
关键词: ADRC;随机控制;库存控制
Keywords: ADRC; stochastic control; inventory control model
基金项目: 国家自然科学基金重点基金资助项目;国家自然科学基金面上基金资助项目;教育部新世纪优秀人才支持计划项目
发表期数: 2017年 第3期
中图分类号: 文献标识码: 文章编号:
参考文献/References:

[1] 韩京清.控制理论——模型论还是控制论[J].系统科学与数学,1989,9(4):328-335.

[2] 韩京清.自抗扰控制技术:估计补偿不确定因素的控制技术[M].北京:国防工业出版社,2008.

[3] Han J. From PID to active disturbance rejection control[J].Industrial Electronics IEEE Transactions on, 2009, 56(3):900-906.

[4] 黄一,张文革.自抗扰控制器的发展[J].控制理论与应用,2002,19(4):485-492.

[5] 韩京清,张荣.二阶扩张状态观测器的误差分析[J].系统科学与数学,1999,(4):465-471.

[6] Holt CC, Modigliani F, Muth JF, et al. Planning production, inventories, and work forces [M].New Jersey: Prentice Hall,1960.

[7] Fleming WH, Sethi SP, Soner HM. An optimal stochastic production planning problem with randomly fluctuating demand [J].SIAM Journal on Control and optimization,1987,25(6):1494-1502.

[8] Sethi SP,Thompson GL. Optimal control theory: applications to management science and economics [M]. Norwell MA: Kluwer,2000.

[9] Dobos,I.Optimal production-inventory strategies for a HMMS-type reverse logistics system [J].International Journal of Production Economics, 2003,81:351-360.

[10] Singhal J, Singhal K. Holt, Modigliani, Muth, and Simon’s work and its role in the renaissance and evolution of operations management[J]. Journal of Operations Management,2007,25:300-309.

[11] Atici,FM, Uysal,F. A production-inventory model of HMMS on time scales[J]. Applied Mathematics Letters,2008,21:236-243.

[12] Dobos I. The analysis of bullwhip effect in a HMMS-type supply chain [J]. International Journal of Production Economics, 2011, 131:250-256.

[13] Dorfman,R. An economic interpretation of optimal control theory[J], American Economic Review,1969,59:817-831.

[14] Gao Z. Active disturbance rejection control: a paradigm shift in feedback control system design[C]. American Control Conference, vol.2006:2399-2405.

[15] Zhang R. An economic interpretation of ADRC[C], Control and Decision Conference (CCDC), 2011 Chinese. IEEE, 2011: 2731-2735.

[16] GUO BZ, ZHAO ZL. On the convergence of an extended state observer for nonlinear systems with uncertainty [J]. Systems & Control Letters,2011,60(6):420-430.

[17] 张荣,韩京清.用模型补偿自抗扰控制器进行参数辨识[J].控制理论及应用,2000,17(1):79-81.