Intelligent Decision Support for Logistics and Supply Chain Management
Permanent URI for this collection
1 - 6 of 6
ItemTowards Predicting Supplier Resilience: A Tree-Based Model Approach( 2022-01-04)ith looming uncertainties and disruptions in today's global supply chains, such as lockdown measures to contain COVID-19, supply chain resilience has gained considerable attention recently. While decision-makers in procurement have emphasized the importance of traditional risk assessment, its shortcomings can be complemented by resilience. However, while most resilience studies are too qualitative in nature and abstract to inform supplier decisions, many quantitative resilience studies frequently rely on complex and impractical operations research models fed with simulated supplier data. Thus there is the need for an integrative, intermediate way for the practical and automated prediction of resilience with real-world data. We therefore propose a random forest-based supervised learning method to predict supplier resilience, outperforming the current human benchmark evaluation by 139 percent. The model is trained on both internal ERP data and publicly available secondary data to help assess suppliers in a pre-screening step, before deciding which supplier to select for a specific product. The results of this study are to be integrated into a software tool developed for measuring and tracking the total cost of supply chain resilience from the perspective of purchasing decisions.
ItemMaritime Spare Parts Management: Current State-of-the-Art( 2022-01-04)Having the right spare part at the right time to the right place for ship maintenance to the minimal possible costs is an exigent management problem that maritime shipping companies face. This is especially challenging in bulk shipping where routes are not fixed, but subsequent port calls depend on spot market dynamics. Thus, spare parts allocation ahead in time is limited, but possible if failures rates of ship components and their timing can be foreseen, so that spare parts can be allocated to hedge against the risk of long waiting times and thus ship downtimes. Thus, monitoring the condition of components key to the ships performance is essential to the task. This can enable companies to significantly reduce operational costs of their fleet leading to a competitive advantage in a highly volatile market regarding demand and demand-driven freight rates. However, shipping companies seem far away from applying such methods due to various challenges ranging from data gathering and cultivating an understanding of data quality needs, adaptation to move from preventive towards predictive and condition-based monitoring, and the introduction and application of decision support tools for sourcing, spare parts allocation, and inventory management. In this paper, we investigate the current state of the art of maintenance and related spare parts logistics management for maritime shipping and discuss the application of methods to the bulk carriage market. We add practical knowledge from case companies and discuss how challenges can be overcome in providing guidelines for companies.
ItemDeep Reinforcement Learning Techniques For Solving Hybrid Flow Shop Scheduling Problems: Proximal Policy Optimization (PPO) and Asynchronous Advantage Actor-Critic (A3C)( 2022-01-04)Well-studied scheduling practices are fundamental for the successful support of core business processes in any manufacturing environment. Particularly, the Hybrid Flow Shop (HFS) scheduling problems are present in many manufacturing environments. The current advances in the field of Deep Reinforcement Learning (DRL) attracted the attention of both practitioners and academics to investigate their adoption beyond synthetic game-like applications. Therefore, we present an approach that is based on DRL techniques in conjunction with a discrete event simulation model to solve a real-world four-stage HFS scheduling problem. The main narrative behind the presented concepts is to expose a DRL agent to a game-like environment using an indirect encoding. Two types of DRL techniques namely, Proximal Policy Optimization (PPO) and Asynchronous Advantage Actor-Critic (A3C), are evaluated for solving problems of different complexity. The computational results suggest that the DRL agents successfully learn appropriate policies for solving the investigated problem. In addition, the investigation shows that the agent can adjust their policies when we expose them to a different problem. We further evaluate the approach to solving problem instances published in the literature to establish a comparison.
ItemArtificial Intelligence in Supply Chain Management: Investigation of Transfer Learning to Improve Demand Forecasting of Intermittent Time Series with Deep Learning( 2022-01-04)Demand forecasting intermittent time series is a challenging business problem. Companies have difficulties in forecasting this particular form of demand pattern. On the one hand, it is characterized by many non-demand periods and therefore classical statistical forecasting algorithms, such as ARIMA, only work to a limited extent. On the other hand, companies often cannot meet the requirements for good forecasting models, such as providing sufficient training data. The recent major advances of artificial intelligence in applications are largely based on transfer learning. In this paper, we investigate whether this method, originating from computer vision, can improve the forecasting quality of intermittent demand time series using deep learning models. Our empirical results show that, in total, transfer learning can reduce the mean square error by 65 percent. We also show that especially short (65 percent reduction) and medium long (91 percent reduction) time series benefit from this approach.
ItemA GPU-Accelerated Approach to Static Stability Assessments for Pallet Loading in Air Cargo( 2022-01-04)The static stability constraint is one of the most important constraints in pallet loading and plays a substantial role when assembling safe and loadable palletizing layouts. Current approaches reach their limits as soon as additional complexity is added, which is a given in the practice of air cargo logistics, or when performance becomes important. As our central objective, we explore a new approach to calculate static stability more performantly and to cover more complexity by relaxing several simplifying assumptions. The approach is implemented in a prototype and builds on the emerging technology of graphical processing unit acceleration in combination with physics engines. We propose a new artifact design and summarize the how-to knowledge in the form of abstracted design principles. Our results demonstrate an improvement in terms of performance depending on the underlying hardware. We develop a conceptual model to assist future research in choosing a solution technology.