Explore the vast range of titles available.

Modern production systems face enormous challenges due to rising customer requirements resulting in complex production systems. The operational efficiency in the competitive industry is ensured by an adequate production control system that manages all operations in order to optimize key performance indicators. Currently, control systems are mostly based on static and model-based heuristics, requiring significant human domain knowledge and, hence, do not match the dynamic environment of manufacturing companies. Data-driven reinforcement learning (RL) showed compelling results in applications such as board and computer games as well as first production applications. This paper addresses the design of RL to create an adaptive production control system by the real-world example of order dispatching in a complex job shop. As RL algorithms are “black box” approaches, they inherently prohibit a comprehensive understanding. Furthermore, the experience with advanced RL algorithms is still limited to single successful applications, which limits the transferability of results. In this paper, we examine the performance of the state, action, and reward function RL design. When analyzing the results, we identify robust RL designs. This makes RL an advantageous control system for highly dynamic and complex production systems, mainly when domain knowledge is limited.

In today’s rapidly changing production landscape with increasingly complex manufacturing processes and shortening product life cycles, a company’s competitiveness depends on its ability to design flexible and resilient production processes. On the shop-floor, in particular, the production control plays a crucial role in coping with disruptions and maintaining system stability and resilience. To address challenges arising from volatile sales markets or other factors, deep learning algorithms have been increasingly applied in production to facilitate fast-paced operations. In particular deep reinforcement learning frequently surpassed conventional and intelligent approaches in terms of performance and computational efficiency and revealed high levels of control adaptability. However, existing approaches were often limited in scope and scenario-specific, which hinders a seamless transition to other control optimization problems. In this paper, we propose a flexible framework that integrates a deep learning based hyper-heuristic into modular production to optimize pre-defined performance indicators. The framework deploys a module recognition and agent experience sharing, enabling a fast initiation of multi-level production systems as well as resilient control strategies. To minimize computational and re-training efforts, a stack of trained policies is utilized to facilitate an efficient reuse of previously trained agents. Benchmark results reveal that our approach outperforms conventional rules in terms of multi-objective optimization. The simulation framework further encourages research in deep-learning-based control approaches to leverage explainability.

EU regulations on CO2 limits and the trend of individualization are pushing the automotive industry towards greater flexibility and robustness in production. One approach to address these challenges is modular production, where workstations are decoupled by automated guided vehicles, requiring new control concepts. Modular production control aims at throughput-optimal coordination of products, workstations, and vehicles. For this np-hard problem, conventional control approaches lack in computing efficiency, do not find optimal solutions, or are not generalizable. In contrast, Deep Reinforcement Learning offers powerful and generalizable algorithms, able to deal with varying environments and high complexity. One of these algorithms is Proximal Policy Optimization, which is used in this article to address modular production control. Experiments in several modular production control settings demonstrate stable, reliable, optimal, and generalizable learning behavior. The agent successfully adapts its strategies with respect to the given problem configuration. We explain how to get to this learning behavior, especially focusing on the agent’s action, state, and reward design.

Continuous product individualization and customization led to the advent of lot size one in production and ultimately to product-inherent uniqueness. As complexities in individualization and processes grow, production systems need to adapt to unique, product-inherent constraints by advancing production control beyond predictive, rigid schedules. While complex processes, production systems and production constraints are not a novelty per se, modern production control approaches fall short of simultaneously regarding the flexibility of complex job shops and product unique constraints imposed on production control. To close this gap, this paper develops a novel, data driven, artificial intelligence based production control approach for complex job shops. For this purpose, product-inherent constraints are resolved by restricting the solution space of the production control according to a prediction based decision model. The approach validation is performed in a real semiconductor fab as a job shop that includes transitional time constraints as product-inherent constraints. Not violating these time constraints is essential to avoid scrap and similarly increase quality-based yield. To that end, transition times are forecasted and the adherence to these product-inherent constraints is evaluated based on one-sided prediction intervals and point estimators. The inclusion of product-inherent constraints leads to significant adherence improvements in the production system as indicated in the real-world semiconductor manufacturing case study and, hence, contributes a novel, data driven approach for production control. As a conclusion, the ability to avoid a large majority of violations of time constraints shows the approaches effectiveness and the future requirement to more accurately integrate such product-inherent constraints into production control.

This paper studies the incentive for vertical information sharing in competing supply chains with production technologies that exhibit diseconomies of scale. We consider a model of two supply chains each consisting of one manufacturer selling to one retailer, with the retailers engaging in Cournot or Bertrand competition. For Cournot retail competition, we show that information sharing benefits a supply chain when (1) the production diseconomy is large and (2) either competition is less intense or at least one retailer's information is less accurate. A supply chain may become worse off when making its information more accurate or production diseconomy smaller, if such an improvement induces the firms in the rival supply chain to cease sharing information. For Bertrand retail competition, we show that information sharing benefits a supply chain when (1) the production diseconomy is large and (2) either competition is less intense or information is more accurate. Under Bertrand competition a manufacturer may be worse off by receiving information, which is never the case under Cournot competition. Information sharing in one supply chain triggers a competitive reaction from the other supply chain and this reaction is damaging to the first supply chain under Cournot competition but may be beneficial under Bertrand competition. This paper was accepted by Martin Lariviere, operations management.

We consider the optimal production and inventory control of an assemble-to-order system with m components, one end-product, and n customer classes. A control policy specifies when to produce each component and, whenever an order is placed, whether or not to satisfy it from on-hand inventory. We formulate the problem as a Markov decision process and characterize the structure of an optimal policy. We show that a base-stock production policy is optimal, but the base-stock level for each component is dynamic and depends on the inventory level of all other components (more specifically, it is nondecreasing). We show that the optimal inventory allocation for each component is a rationing policy with different rationing levels for different demand classes. The rationing levels for each component are dynamic and also nondecreasing in the inventory level of all other components. We compare the performance of the optimal policy to heuristic policies, including the commonly used base-stock policy with fixed base-stock levels, and find them to perform surprisingly well, especially for systems with lost sales.

Production control systems (PCSs) control the flow of jobs in a production system. The selection of a suitable PCS in the context of make-to-order (MTO) is challenging, due to the characteristics of MTO businesses and the number of parameters or factors that comprise a PCS. The literature that compares PCSs in the MTO context reported contradictory results. In fact, there is a gap in the literature concerning which factors or parameters explain a PCS performance. This paper presents an analysis of comparative studies on PCS in the MTO context, using a systematic literature review, to reveal which control factors and manufacturing conditions influence a PCS performance. The analysis concentrates on studies that use simulation to assess the performance of PCSs. Our results indicate that the main difference in PCSs performance is the design of the control loops. Other important factors that must be considered in the choice of a PCS are the order release mechanism, the workload aggregation approach, and the workload estimation method used on control loops. A framework for choosing a suitable PCS for MTO companies is presented, considering these factors.

The subsea production system is a vital equipment for offshore oil and gas production. The control system is one of the most important parts of it. Collecting and processing the signals of subsea sensors is the only way to judge whether the subsea production control system is normal. However, subsea sensors degrade rapidly due to harsh working environments and long service time. This leads to frequent false alarm incidents. A combinatorial reasoning-based abnormal sensor recognition method for subsea production control system is proposed. A combinatorial algorithm is proposed to group sensors. The long short-term memory network (LSTM) is used to establish a single inference model. A counting-based judging method is proposed to identify abnormal sensors. Field data from an offshore platform in the South China Sea is used to demonstrate the effect of the proposed method. The results show that the proposed method can identify the abnormal sensors effectively.

In many applications involving make-to-order or time-sensitive (e.g., perishable, seasonal) products, finished orders are often delivered to customers immediately or shortly after the production. Consequently, there is little or no finished product inventory in the supply chain such that production and outbound distribution are very intimately linked and must be scheduled jointly to achieve a desired on-time delivery performance at minimum total cost. Research on integrated scheduling models of production and outbound distribution is relatively recent but is growing very rapidly. In this paper, we provide a survey of such existing models. We present a unified model representation scheme, classify existing models into several different classes, and for each class of the models give an overview of the optimality properties, computational tractability, and solution algorithms for the various problems studied in the literature. We clarify the tractability of some open problems left in the literature and some new problems by providing intractability proofs or polynomial-time exact algorithms. We also identify several problem areas and issues for future research.

In actual dynamic system, uncertainty is absolute and certainty is relative. This paper presents the optimal dynamic pricing and production control strategy for perishable products in finite horizon. The influence of external environmental disturbance on the system is considered by means of a special uncertain process (Liu process). Then based on uncertainty theory and Hurwicz criterion, the optimization model is built, where control variables are restricted to an admissible control set. In addition, uncertain differential equation is used to describe the changes of inventory. By applying the optimality equation, we determine the optimal price and production strategy to maximize profit. Besides, both the optimal price and production rate are linearly decreasing with inventory. Afterwards, two numerical examples are given, the results reveal that reducing the uncertain disturbance of inventory and expanding the potential market size are beneficial to improving the optimal profit. Moreover, risk-loving decision makers can gain more profits while facing large risks.