Adaptive inventory control models for supply chain management

Uncertainties inherent in customer demands make it difficult for supply chains to achieve just-in-time inventory replenishment, resulting in loosing sales opportunities or keeping excessive chain-wide inventories. In this paper, we propose two adaptive inventory-control models for a supply chain consisting of one supplier and multiple retailers. The one is a centralized model and the other is a decentralized model. The objective of the two models is to satisfy a target service level predefined for each retailer. The inventory-control parameters of the supplier and retailers are safety lead time and safety stocks, respectively. Unlike most extant inventory-control approaches, modelling the uncertainty of customer demand as a statistical distribution is not a prerequisite in the two models. Instead, using a reinforcement learning technique called action-value method, the control parameters are designed to adaptively change as customer-demand patterns changes. A simulation-based experiment was performed to compare the performance of the two inventory-control models.     

基于集成学习算法的生物质气化炉改进模糊双闭环控制系统

生物质气化过程是具有弱随机性的物理化学过程,针对其工艺特点本文提出一种基于集成预测和改进模糊的控制算法.外环采用基于SVM的AdaBoost集成学习方法,预测温度变化趋势;内环采用了遗传模糊算法,将预测结果和含氧量反馈值作为输入,在稳定炉顶层温度和降低含氧量这对矛盾之间,寻求最优鼓风机的输出,对一次风进风量进行调节.仿真和现场运行数据证明了方法够满足气化炉的工艺要求和控制目标.     

A new fuzzy network data envelopment analysis model for measuring the performance of agility in supply chains

Data envelopment analysis (DEA) is a linear programming method for assessing the efficiency and productivity of organizational units called decision-making units (DMUs). We propose a new network DEA (NDEA) model for measuring the performance of agility in supply chains. The uncertainty of the input and output data is modeled with linguistic terms parameterized with fuzzy sets. The proposed fuzzy NDEA model is linear and independent of the α-cut variables. The linear feature allows for a quick identification of the global optimum solution and the α-cut independency feature allows for a significant reduction in the computational efforts. We show that our model always generate solutions within a bounded feasible region. Our model also eliminates the potential for conflict by producing unique interval efficiency scores for each DMU. The proposed model is used to measure the performance of agility in a real-life case study in the dairy industry.     

Tuning fuzzy PD and PI controllers using reinforcement learning

In this paper, we propose a new auto-tuning fuzzy PD and PI controllers using reinforcement Q-learning (QL) algorithm for SISO (single-input single-output) and TITO (two-input two-output) systems. We first, investigate the design parameters and settings of a typical class of Fuzzy PD (FPD) and Fuzzy PI (FPI) controllers: zero-order Takagi–Sugeno controllers with equidistant triangular membership functions for inputs, equidistant singleton membership functions for output, Larsen’s implication method, and average sum defuzzification method. Secondly, the analytical structures of these typical fuzzy PD and PI controllers are compared to their classical counterpart PD and PI controllers. Finally, the effectiveness of the proposed method is proven through simulation examples.     

双足机器人步态控制的深度强化学习方法

针对双足机器人行走过程中的步态稳定控制问题,提出一种改进深度Q网络的深度强化学习方法.首先,将深度Q网络算法与确定性策略梯度相结合,提出用修正Double-Q网络优化操作一评论网络的评论网络,给出一种改进的深度Q网络;然后,建立双足机器人连杆模型,在常规的平整路面上将改进的深度Q网络用于作为智能体的双足机器人进行步态控...     

Optimal control in microgrid using multi-agent reinforcement learning

This paper presents an improved reinforcement learning method to minimize electricity costs on the premise of satisfying the power balance and generation limit of units in a microgrid with grid-connected mode. Firstly, the microgrid control requirements are analyzed and the objective function of optimal control for microgrid is proposed. Then, a state variable "Average Electricity Price Trend" which is used to express the most possible transitions of the system is developed so as to reduce the complexity and randomicity of the microgrid, and a multi-agent architecture including agents, state variables, action variables and reward function is formulated. Furthermore, dynamic hierarchical reinforcement learning, based on change rate of key state variable, is established to carry out optimal policy exploration. The analysis shows that the proposed method is beneficial to handle the problem of "curse of dimensionality" and speed up learning in the unknown large-scale world. Finally, the simulation results under JADE (Java Agent Development Framework) demonstrate the validity of the presented method in optimal control for a microgrid with grid-connected mode.     

跨链间同级库存协作下集群式供应链协调分析

在跨链间同级库存协作机制下,采用收益共享契约研究了多链式集群供应链系统的协调问题。建立了零售商跨链间库存协作机制下的集群式供应链系统分散决策模型、集中决策模型以及收益共享契约模型,并给出各模型下的最优策略。通过数值仿真对模型中的参数进行求解,并对结果进行了分析和验证。研究结果表明,在跨链间同级库存协作机制下,"双重边际化"效应依然存在于集群式供应链系统内,但收益共享契约的引入可以有效提高系统内各成员的利润,并实现集群式供应链系统的整体协调;同时收益分配系数的取值范围与需求不确定性和跨链间同级库存协作成员数量分别存在逆相关性和正相关性,而收益共享契约引入的必要性与二者均存在正相关性。     

恒压供水系统的模糊控制

依据现场操作人员的控制经验，设计一种由PLC和变频器实现的模糊恒压供水控制系统，克服了传统PID控制设计中的参数调整困难的问题，该系统取代了高位水塔和直接水泵加压供水方式，具有明显的节能效果。     

基于元胞机与强化学习的多扰动车间调度算法

针对大型装备制造企业扰动多、影响大的问题,以元胞机模型为框架构建了多扰动车间生产调度模型,设计了基于设备平均利用率与工件平均流程时间双目标最优的目标函数,并通过算例验证了模型的科学性.同时,考虑到多扰动车间调度的复杂性,为寻找全局最优解,采用强化学习算法优化了元胞机的自组织演化规则,提出了针对设备故障、紧急插单与新订单干扰三种典型扰动的调度策略,最终建立了基于元胞机与强化学习算法的多扰动车间柔性调度模型.以某大型零件制造企业为例,说明了模型的具体优化过程,并通过仿真求解验证了算法与模型的有效性与可靠性.     

Two-layer networked learning control using self-learning fuzzy control algorithms

Since the existing single-layer networked control systems have some inherent limitations and cannot effectively handle the problems associated with unreliable networks, a novel two-layer networked learning control system （NLCS） is proposed in this paper. Its lower layer has a number of local controllers that are operated independently, and its upper layer has a learning agent that communicates with the independent local controllers in the lower layer. To implement such a system, a packet-discard strategy is firstly developed to deal with network-induced delay and data packet loss. A cubic spline interpolator is then employed to compensate the lost data. Finally, the output of the learning agent based on a novel radial basis function neural network （RBFNN） is used to update the parameters of fuzzy controllers. A nonlinear heating, ventilation and air-conditioning （HVAC） system is used to demonstrate the feasibility and effectiveness of the proposed system.     

基于拉格朗日松弛和遗传算法的供应链协同计划

研究了多级供应链伙伴间的生产计划协调优化问题。对集成性多阶段约束生产批量计划模型增加了关联约束和相关需求约束。基于模型的加可分性结构,利用拉格朗日松弛技术将其分解为成员独立的子问题。应用遗传算法更新拉格朗日乘子来协调成员决策,在不干涉成员决策权和私有信息的前提下,实现了多级供应链生产计划的协调优化。仿真实验证明了基于拉格朗日松弛技术与遗传算法的计划协同模式和协调方法的优越性和鲁棒性。     

Impedance control method with reinforcement learning for dual-arm robot installing slabstone

Journal of Mechanical Science and Technology - Slabstone installation task in the construction field is associated with large size, heavy weight, and unstructured environment, making it necessary...     

A computer-integrated FMEA for dynamic supply chains in a flexible-based environment

Quality management (QM) for dynamic supply chains (DSCs) in a flexible-based environment has received increasing attention in current business environment. This study focuses on the development of an QM tool, based on the failure modes and effects analysis (FMEA) approach, and demonstrates how it can be extended from application in supply chain management (SCM) to other business environments. The main contributions are to implement an iterative process using FMEA for DSCs in a flexible-based environment by overcoming the weaknesses associated with traditional FMEA system and to develop a user-friendly interface, with which the distributed parties in chains could cooperate as a whole and meet the needs of dynamic control. In specific, this paper proposes a computer-integrated FMEA approach that enhanced FMEA in SCM through automated processing using a fuzzy approach and a computer-integrated and internet-based interface to support the system implementation. The efficiency and practical implication of the proposed system has been validated by the developed system with a case study. The developed system can help prevent possible failures in the design and operations and, furthermore, the assurance of quality for DSCs in a flexible-based environment. The proposed system may easily be extended to more complex real-world applications.     

Supplier selection problem in global supply chains by AHP and ANP approaches under fuzzy environment

In this paper, we proposed a model for selecting of the global supplier by analytical hierarchy process (AHP) and analytical network process (ANP) based on linguistic variable weight. The fuzzy AHP and fuzzy ANP methods are suggested to be a good tool for solving our proposed multiple-criteria decision-making problem. Fuzzy AHP-based methodology evaluated different decision criteria such as service quality, cost, risk factors, and supplier’s characteristics involved in the selection of the best supplier in a global supply chain and ANP represented an effective tool for providing a suitable solution for managers. The linguistic levels of comparisons produced by the experts with triangular fuzzy numbers for constructing fuzzy pairwise comparison matrices. The implementation of the system was demonstrated by a problem having four stages of hierarchy which contains four criteria and 28 attributes. In conclusion, the fuzzy AHP and fuzzy ANP results were compared.     

采用多智能体强化学习的交通信号优化控制

在城市交通环境,交通流的正确预测是比较困难,因为多个十字路口,这使得预设的交通控制模型之间的相互作用和纠缠在一起,不能在所有的交通情况下始终保持高性能的预测。考虑到的强化学习的所具有的自主学习能力,本文提出了基于多智能体强化学习的交通信号控制方法。没有预设的控制模型,多协作代理可以学习相应的实时交通状况下的最优控制策略。通过实验结果证明了这种方法的可行性和有效性。     

A polynomial time algorithm for production planning with bounded inventory

In petrochemical manufacturing, food processing, glass manufacturing, etc., there is a situation where the production planning is subject to the limit of inventory capacity (or bounded inventory) rather than production capacity. Only a few studies can be found in the literature for this situation. In this paper, the production planning problem with bounded inventory is discussed. Specifically, the problem considered had the following features: (i) backlogging is not allowed, (ii) lost sale is allowed, (iii) the lost sale cost function is time varying and non-increasing, and (iv) inventory capacity is non-decreasing. An algorithm for this problem with the polynomial O(T ³) time complexity is proposed. Though the complexity of the algorithm is the same as that of the other work, see Sandbothe and Thompson (Oper Res 38(3):474–486, 1990), the problem in the other’s work problem is less complex than ours; for instance only a fixed cost function is considered. A case study taken from a real-life production line is provided to further justify the application of the lost sale strategy, demonstrating the effectiveness of our model and algorithm.     

集合标志Petri网的建模方法及在供应链库存控制中的应用

为解决供应链库存系统控制问题,提出了一种集合标志Petri网建模与分析方法。首先扩展集合标志及其属性来描述多品种实体群组或相应的信息。在语义上扩展了结构元素和操作运算与函数。结构和运行规则的扩展规定了如何构建模型和模型如何自执行。采用AutoMod完成了集合标志Petri网的动态执行与仿真分析。改进了一个3层次．4节点、一百多品种的供应链库存系统的时序机制设计,提出了考虑装配型特点的综合分类方法,弗结合信息共享和响应机制的改进,用集合标志Petri网求得整体控制方案参数的满意解。结果显示,供应链库存系统的整体库存显著下降,周转明显加快,服务率进一步提高。     

A two-stage data envelopment analysis model for measuring performance in three-level supply chains

Data Envelopment Analysis (DEA) has been widely used to evaluate supply chain performances. In conventional DEA, supply chains are represented as black boxes where only the initial inputs and final outputs are considered to measure their efficiency. However, an integrated model measuring both the efficiency of the entire supply chain and that of all its components at all levels is essential for a comprehensive evaluation. This study presents a two-stage DEA method to evaluate the performance of a three-level supply chain including suppliers, manufacturers and distributors. The proposed model can be used both under the constant returns to scale and the variable returns to scale assumptions and can be easily implemented for comprehensive analysis of multi-level supply chains. We present a numerical example to demonstrate applicability of the proposed model and exhibit the efficacy and effectiveness of the proposed algorithms and procedures. In particular, the numerical results demonstrate that the entire supply chain is “comprehensively” efficient only if efficient supplier–manufacturer and manufacturer–distributor relationships are established.     

供应链管理中的战略库存

从库存的角度分析供应链的整体优化问题。首先探讨了供应链的几种结构及其应的解耦库存的位置,分析了“推”式供应链与“拉”式供应链的主要区别;然后运用约束理论分析供应链中缓冲库存的设置。     

A reinforcement learning approach for developing routing policies in multi-agent production scheduling

Most recent research studies on agent-based production scheduling have focused on developing negotiation schema for agent cooperation. However, successful implementation of agent-based approaches not only relies on the cooperation among the agents, but the individual agent’s intelligence for making good decisions. Learning is one mechanism that could provide the ability for an agent to increase its intelligence while in operation. This paper presents a study examining the implementation of the Q-learning algorithm, one of the most widely used reinforcement learning approaches, for use by job agents when making routing decisions in a job shop environment. A factorial experiment design for studying the settings used to apply Q-learning to the job routing problem is carried out. This study not only investigates the effects of this Q-learning application but also provides recommendations for factor settings and useful guidelines for future applications of Q-learning to agent-based production scheduling.