In recent years, supply chains have become increasingly globalized. As a consequence, the world's supply of all types of parts has become more susceptible to disruptions. Some of these disruptions are extreme and ...
详细信息
In recent years, supply chains have become increasingly globalized. As a consequence, the world's supply of all types of parts has become more susceptible to disruptions. Some of these disruptions are extreme and may have global implications. Our research is based on the supply risk management problem faced by a manufacturer. We model the problem as a dynamic program, design and implement approximate dynamic programming (ADP) algorithms to solve it, to overcome the well-known curses of dimensionality. Using numerical experiments, we compare the performance of different ADP algorithms. We then design a series of numerical experiments to study the performance of different sourcing strategies (single, dual, multiple, and contingent sourcing) under various settings, and to discover insights for supply risk management practice. The results show that, under a wide variety of settings, the addition of a third or more suppliers brings much less marginal benefits. Thus, managers can limit their options to a backup supplier (contingent sourcing) or an additional regular supplier (dual sourcing). Our results also show that, unless the backup supplier can supply with zero lead time, using dual sourcing appears to be preferable. Lastly, we demonstrate the capability of the proposed method in analyzing more complicated realistic supply chains. (C) 2012 Elsevier Ltd. All rights reserved.
This paper illustrates the development of an intelligent local area signals based controller for damping low-frequency oscillations in power systems. The controller is trained offline to perform well under a wide vari...
详细信息
This paper illustrates the development of an intelligent local area signals based controller for damping low-frequency oscillations in power systems. The controller is trained offline to perform well under a wide variety of power system operating points, allowing it to handle the complex, stochastic, and time-varying nature of power systems. Neural network based system identification eliminates the need to develop accurate models from first principles for control design, resulting in a methodology that is completely data driven. The virtual generator concept is used to generate simplified representations of the power system online using time-synchronized signals from phasor measurement units at generating stations within an area of the system. These representations improve scalability by reducing the complexity of the system "seen" by the controller and by allowing it to treat a group of several synchronous machines at distant locations from each other as a single unit for damping control purposes. A reinforcement learning mechanism for approximate dynamic programming allows the controller to approach optimality as it gains experience through interactions with simulations of the system. Results obtained on the 68-bus New England/New York benchmark system demonstrate the effectiveness of the method in damping low-frequency inter-area oscillations without additional control effort.
The objective of infrastructure management is to provide optimal maintenance, rehabilitation and replacement (MR&R) policies for a system of facilities over a planning horizon. While most approaches in the literat...
详细信息
The objective of infrastructure management is to provide optimal maintenance, rehabilitation and replacement (MR&R) policies for a system of facilities over a planning horizon. While most approaches in the literature have studied the decision-making process as a finite resource allocation problem, the impact of construction activities on the road network is often not accounted for. The state-of-the-art Markov decision process (MDP)-based optimization approaches in infrastructure management, while optimal for solving budget allocation problems, become internally inconsistent upon introducing network constraints. In comparison, approximate dynamic programming (ADP) enables solving complex problem formulations by using simulation techniques and lower dimension value function approximations. In this paper, an ADP framework is proposed, wherein capacity losses due to construction activities are subjected to an agency-defined network capacity threshold. A parametric study is conducted on a stylized network configuration to infer the impact of network-based constraints on the decision-making process. (C) 2013 Elsevier Ltd. All rights reserved.
approximate dynamic programming is a useful tool in solving multi-stage decision optimal control problems. In this work, we first promote the action-dependent heuristic dynamicprogramming method to multi-input multi-...
详细信息
approximate dynamic programming is a useful tool in solving multi-stage decision optimal control problems. In this work, we first promote the action-dependent heuristic dynamicprogramming method to multi-input multi-output control system by extending its action network to multi-output form. The detailed derivation is also given. We then apply this method to the fluctuation control of a spark engine idle speed. An engine idling model is set up to verify the control effect of this method. Results here show that this method requires several iterations to suppress unbalanced combustion by manipulating spark ignition timing. This method provides an alternative for a simpler multi-input multi-output approximate dynamic programming scheme. Moreover, it has a faster iteration convergence effect. The derivation of this method also has a rigorous mathematical basis. Although illustrated for engines, this control system framework should also be applicable to general multi-input multi-output nonlinear system. Copyright (C) 2011 John Wiley & Sons, Ltd.
Efficient proactive emergency vehicle planning is crucial for effective emergency response management systems. This paper develops a flexible deep reinforcement learning framework to address uncertainties in dynamic e...
详细信息
Efficient proactive emergency vehicle planning is crucial for effective emergency response management systems. This paper develops a flexible deep reinforcement learning framework to address uncertainties in dynamic emergency requests and complex traffic conditions. The Markov decision process model is proposed to optimize ambulance dispatch and reallocation, focusing on minimizing average response time, reducing delays from traffic congestion and patient severity misclassification, and enhancing general fairness. To tackle the challenges of the unbounded MDP state space, we employ a least-squares-based approximate policy iteration model for the upper bound and a linear programming model for the lower bound. We use Chinese emergency medical services system data for computational experiments and select average response time, fraction of delay, risk level, and Gini coefficient to evaluate the performance of our model. The numerical results demonstrate our optimal flexible policy outperforms the first-in-first-served rule, improving both efficiency and equity in medical decision-making.
This paper addresses a real-life stochastic car batching and sequencing problem in the body shop of a vehicle plant. Unlike previous research on similar problems in the paint shop or assembly shop, our problem primari...
详细信息
This paper addresses a real-life stochastic car batching and sequencing problem in the body shop of a vehicle plant. Unlike previous research on similar problems in the paint shop or assembly shop, our problem primarily focuses on production planning in the body shop, with specific constraints on the production sequence (e.g., the bodies of each model cannot be produced individually but must be produced in batches of a specific size) and considers the impact of sampling inspection. Solving this large-scale car batching and sequencing problem in the body shop within an acceptable computation time is challenging. In this paper, several efficient dynamicprogramming-based algorithms are designed to solve the problem. First, a dynamicprogramming model is established for the deterministic version of the problem, and the optimal solution can be obtained by the dynamicprogramming approach. Furthermore, faced with the uncertainty introduced by sampling inspection, a more difficult and practical stochastic car batching and sequencing problem is modeled as a discrete-time Markov decision process. A rollout method-based approximate dynamic programming algorithm is designed to solve this complex problem. Finally, the proposed algorithms' effectiveness is examined using real-life production data. Note to Practitioners-This paper is motivated by our collaboration with a vehicle production plant in Shanghai, China. The plant mainly produces 3 similar to 4 electrified models, with an annual output of up to 300,000 units/year. The plant consists of three workshops: the body shop, the paint shop, and the assembly shop. A multi-model flexible production line has been set up in the body shop. In practice, as multiple models are produced on the production line simultaneously, the plant adopts a mode of batch production to batch and sequence the bodies to be produced. Due to the specialty of the batch production mode and the uncertainty brought by the sampling inspection, the stochasti
Container slot allocation represents a critical operational decision -making challenge within the liner shipping industry, which necessitates making decisions on the transportation of loaded containers and the reposit...
详细信息
Container slot allocation represents a critical operational decision -making challenge within the liner shipping industry, which necessitates making decisions on the transportation of loaded containers and the repositioning of empty ones under stochastic demand. We study novel dynamic allocation policy that leverage sequentially -revealed demand information to determine the slot allocation decision for both loaded and empty containers at each stage. In this paper, we develop a stochastic dynamicprogramming (DP) model to optimize the slot allocation decision for maximizing the expected total revenue over the planning horizon. To solve this model, we design an efficient allocation policy that makes slot allocations at each stage based on current empty container stocks, realized demand, and the mean demand at future stages. Comprehensive numerical experiments on both synthetic and realistic data demonstrate substantial revenue improvement of our approach over the commonly -used benchmark policies in practice and literature.
We explore a problem faced by agri-food e-commerce platforms in purchasing different, perishable products and collecting them from multiple producers and delivering them to a single warehouse, aiming to maintain adequ...
详细信息
We explore a problem faced by agri-food e-commerce platforms in purchasing different, perishable products and collecting them from multiple producers and delivering them to a single warehouse, aiming to maintain adequate inventory levels to meet current and future customer demand, while avoiding waste. Customer demand and suppliers' purchase prices and supply volumes are uncertain and revealed on a periodical basis. Every period, purchasing, inventory, and routing decisions are made to satisfy demand and to build inventory for future periods. For effective decisions integrating all three decision components and anticipating future developments, we propose a stochastic lookahead method that, in every period, samples future scenarios for demand, supply volumes, and prices. It then solves a two-stage stochastic program to obtain the decision for the current period. To make this approach computationally tractable, we reduce the routing decision in the two-stage program and use an approximate routing cost instead. Given the reduced decision, we then create the final decision via a conventional routing heuristic. We learn the routing cost approximation adaptively via repeated training simulations. In comprehensive experiments, we show that all three components, stochastic lookahead, routing cost approximation, and adaptive learning, are very effective individually, but especially in combination. We also provide a comprehensive analysis of the problem parameters and obtain valuable insights in problem and methodology.
In social networks, the influence maximization problem requires selecting an initial set of nodes to influence so that the spread of influence can reach its maximum under certain diffusion models. Usually, the problem...
详细信息
In social networks, the influence maximization problem requires selecting an initial set of nodes to influence so that the spread of influence can reach its maximum under certain diffusion models. Usually, the problem is formulated in a two-stage un-budgeted fashion: The decision maker selects a given number of nodes to influence and observes the results. In the adaptive version of the problem, it is possible to select the nodes at each time step of a given time interval. This allows the decision-maker to exploit the observation of the propagation and to make better decisions. This paper considers the adaptive budgeted influence maximization problem, that is, the adaptive problem in which the decision maker has a finite budget to influence the nodes, and each node requires a cost to be influenced. We present two solution techniques: The first is an approximated value iteration leveraging mixed integer linear problems while the second exploits new concepts from graph neural networks. Extensive numerical experiments demonstrate the effectiveness of the proposed approaches.
This paper investigates the approximate optimal coordination for nonlinear uncertain second -order multirobot systems with guaranteed safety (collision avoidance) Through constructing novel local error signals, the co...
详细信息
This paper investigates the approximate optimal coordination for nonlinear uncertain second -order multirobot systems with guaranteed safety (collision avoidance) Through constructing novel local error signals, the collision -free control objective is formulated into an coordination optimization problem for nominal multirobot systems. Based on approximate dynamic programming technique, the optimal value functions and control policies are learned by simplified critic -only neural networks (NNs). Then, the approximated optimal controllers are redesigned using adaptive law to handle the effects of robots' uncertain dynamics. It is shown that the NN weights estimation errors are uniformly ultimately bounded under proper conditions, and safe coordination of multiple robots can be achieved regardless of model uncertainties. Numerical simulations finally illustrate the effectiveness of the proposed controller.
暂无评论