Lotte van Hezewijk
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Lotte van Hezewijk has a total of 10 co-authors. Here are the top co-authors by number of shared publications:
Nico P. Dellaert
2 shared publications- Kevin Geevers
ORTEC B.V., Zoetermeer, The Netherlands
1 shared publication - Martijn Mes
Department of High-Tech Business and Entrepreneurship , University of Twente , Enschede , 7500 AE , Netherlands , utwente.nl
1 shared publication Nico Dellaert
Industrial Engineering & Innovation Sciences Eindhoven University of Technology P.O. Box 513 5600 MB Eindhoven The Netherlands
1 shared publication- Nico Dellaert
School of Industrial Engineering, Eindhoven University of Technology, 5600MB Eindhoven, Netherlands
1 shared publication - Noud Gademann
ORTEC B.V., Zoetermeer, Netherlands
1 shared publication
Latest Publications
- Open Access
- 26 March 2025
- Published by Elsevier in International Journal of Production Economics
- Vol. 284
Scalable deep reinforcement learning in the non-stationary capacitated lot sizing problem
- 26 March 2025
- Published by Elsevier in International Journal of Production Economics
- Vol. 284
Open AccessScalable deep reinforcement learning in the non-stationary capacitated lot sizing problem
- Open Access
- 6 March 2025
- Published by Springer Nature in Mathematical Methods of Operations Research
- Vol. 101 (2) , 135-161
On non-negative auto-correlated integer demand processes
Methods to generate realistic non-stationary demand scenarios are a key component for analyzing and optimizing decision policies in supply chains. Typical forecasting techniques recommended in standard inventory control textbooks consist of some form of simple exponential smoothing (SES) for both the estimates for the mean and standard deviation. We study demand generating processes (DGPs) that yield non-stationary demand scenarios, and that are consistent with SES, meaning that SES yields unbiased estimates when applied to the generated demand scenarios. As demand in typical practical settings is discrete and non-negative, we study consistent DGPs on the non-negative integers. We derive conditions under which the existence of such DGPs can be guaranteed, and propose a specific DGP that yields autocorrelated, discrete demands when these conditions are satisfied. Our subsequent simulation study gains further insights into the proposed DGP. It demonstrates that from a given initial forecast, our DGP yields a diverse set of demand scenarios with a wide range of properties. To show the applicability of the DGP, we apply it to generate demand in a standard inventory problem with full backlogging and a positive lead time. We find that appropriate dynamic base-stock levels can be obtained using a new and relatively simple algorithm, and we demonstrate that this algorithm outperforms relevant benchmarks.
- 6 March 2025
- Published by Springer Nature in Mathematical Methods of Operations Research
- Vol. 101 (2) , 135-161
Open AccessOn non-negative auto-correlated integer demand processes
Methods to generate realistic non-stationary demand scenarios are a key component for analyzing and optimizing decision policies in supply chains. Typical forecasting techniques recommended in standard inventory control textbooks consist of some form of simple exponential smoothing (SES) for both the estimates for the mean and standard deviation. We study demand generating processes (DGPs) that yield non-stationary demand scenarios, and that are consistent with SES, meaning that SES yields unbiased estimates when applied to the generated demand scenarios. As demand in typical practical settings is discrete and non-negative, we study consistent DGPs on the non-negative integers. We derive conditions under which the existence of such DGPs can be guaranteed, and propose a specific DGP that yields autocorrelated, discrete demands when these conditions are satisfied. Our subsequent simulation study gains further insights into the proposed DGP. It demonstrates that from a given initial forecast, our DGP yields a diverse set of demand scenarios with a wide range of properties. To show the applicability of the DGP, we apply it to generate demand in a standard inventory problem with full backlogging and a positive lead time. We find that appropriate dynamic base-stock levels can be obtained using a new and relatively simple algorithm, and we demonstrate that this algorithm outperforms relevant benchmarks.
- Open Access
On non-negative auto-correlated integer demand processes
Methods to generate realistic non-stationary demand scenarios are a key component for analyzing and optimizing decision policies in supply chains. Typical forecasting techniques recommended in standard inventory control textbooks consist of some form of simple exponential smoothing (SES) for both the estimates for the mean and standard deviation. We study demand generating processes (DGPs) that yield non-stationary demand scenarios, and that are consistent with SES, meaning that SES yields unbiased estimates when applied to the generated demand scenarios. As demand in typical practical settings is discrete and non-negative, we study consistent DGPs on the non-negative integers. We derive conditions under which the existence of such DGPs can be guaranteed, and propose a specific DGP that yields autocorrelated, discrete demands when these conditions are satisfied. Our subsequent simulation study gains further insights into the proposed DGP. It demonstrates that from a given initial forecast, our DGPs yields a diverse set of demand scenarios with a wide range of properties. To show the applicability of the DGP, we apply it to generate demand in a standard inventory problem with full backlogging and a positive lead time. We find that appropriate dynamic base-stock levels can be obtained using a new and relatively simple algorithm, and we demonstrate that this algorithm outperforms relevant benchmarks.
Open AccessOn non-negative auto-correlated integer demand processes
Methods to generate realistic non-stationary demand scenarios are a key component for analyzing and optimizing decision policies in supply chains. Typical forecasting techniques recommended in standard inventory control textbooks consist of some form of simple exponential smoothing (SES) for both the estimates for the mean and standard deviation. We study demand generating processes (DGPs) that yield non-stationary demand scenarios, and that are consistent with SES, meaning that SES yields unbiased estimates when applied to the generated demand scenarios. As demand in typical practical settings is discrete and non-negative, we study consistent DGPs on the non-negative integers. We derive conditions under which the existence of such DGPs can be guaranteed, and propose a specific DGP that yields autocorrelated, discrete demands when these conditions are satisfied. Our subsequent simulation study gains further insights into the proposed DGP. It demonstrates that from a given initial forecast, our DGPs yields a diverse set of demand scenarios with a wide range of properties. To show the applicability of the DGP, we apply it to generate demand in a standard inventory problem with full backlogging and a positive lead time. We find that appropriate dynamic base-stock levels can be obtained using a new and relatively simple algorithm, and we demonstrate that this algorithm outperforms relevant benchmarks.
- Open Access
- 19 July 2023
- Published by Springer Nature in Central European Journal of Operations Research
- Vol. 32 (3) , 653-683
Multi-echelon inventory optimization using deep reinforcement learning
This paper studies the applicability of a deep reinforcement learning approach to three different multi-echelon inventory systems, with the objective of minimizing the holding and backorder costs. First, we conduct an extensive literature review to map the current applications of reinforcement learning in multi-echelon inventory systems. Next, we apply our deep reinforcement learning method to three cases with different network structures (linear, divergent, and general structures). The linear and divergent cases are derived from literature, whereas the general case is based on a real-life manufacturer. We apply the proximal policy optimization (PPO) algorithm, with a continuous action space, and show that it consistently outperforms the benchmark solution. It achieves an average improvement of 16.4% for the linear case, 11.3% for the divergent case, and 6.6% for the general case. We explain the limitations of our approach and propose avenues for future research.
- 19 July 2023
- Published by Springer Nature in Central European Journal of Operations Research
- Vol. 32 (3) , 653-683
Open AccessMulti-echelon inventory optimization using deep reinforcement learning
This paper studies the applicability of a deep reinforcement learning approach to three different multi-echelon inventory systems, with the objective of minimizing the holding and backorder costs. First, we conduct an extensive literature review to map the current applications of reinforcement learning in multi-echelon inventory systems. Next, we apply our deep reinforcement learning method to three cases with different network structures (linear, divergent, and general structures). The linear and divergent cases are derived from literature, whereas the general case is based on a real-life manufacturer. We apply the proximal policy optimization (PPO) algorithm, with a continuous action space, and show that it consistently outperforms the benchmark solution. It achieves an average improvement of 16.4% for the linear case, 11.3% for the divergent case, and 6.6% for the general case. We explain the limitations of our approach and propose avenues for future research.
- Open Access
- 7 April 2022
- Published by Taylor & Francis in International Journal of Production Research
- Vol. 61 (6) , 1955-1978
Using the proximal policy optimisation algorithm for solving the stochastic capacitated lot sizing problem
This paper studies the multi-item stochastic capacitated lot-sizing problem with stationary demand to minimise set-up, holding, and backorder costs. This is a common problem in the industry, concerning both inventory management and production planning. We study the applicability of the Proximal Policy Optimisation (PPO) algorithm in this problem, which is a type of Deep Reinforcement Learning (DRL). The problem is modelled as a Markov Decision Process (MDP), which can be solved to optimality in small problem instances by using Dynamic Programming. In these settings, we show that the performance of PPO approaches the optimal solution. For larger problem instances with an increasing number of products, solving to optimality is intractable, and we demonstrate that the PPO solution outperforms the benchmark solution. Several adjustments to the standard PPO algorithm are implemented to make it more scalable to larger problem instances. We show the linear growth in computation time for the algorithm, and present a method for explaining the outcomes of the algorithm. We suggest future research directions that could improve the scalability and explainability of the PPO algorithm.
- 7 April 2022
- Published by Taylor & Francis in International Journal of Production Research
- Vol. 61 (6) , 1955-1978
Open AccessUsing the proximal policy optimisation algorithm for solving the stochastic capacitated lot sizing problem
This paper studies the multi-item stochastic capacitated lot-sizing problem with stationary demand to minimise set-up, holding, and backorder costs. This is a common problem in the industry, concerning both inventory management and production planning. We study the applicability of the Proximal Policy Optimisation (PPO) algorithm in this problem, which is a type of Deep Reinforcement Learning (DRL). The problem is modelled as a Markov Decision Process (MDP), which can be solved to optimality in small problem instances by using Dynamic Programming. In these settings, we show that the performance of PPO approaches the optimal solution. For larger problem instances with an increasing number of products, solving to optimality is intractable, and we demonstrate that the PPO solution outperforms the benchmark solution. Several adjustments to the standard PPO algorithm are implemented to make it more scalable to larger problem instances. We show the linear growth in computation time for the algorithm, and present a method for explaining the outcomes of the algorithm. We suggest future research directions that could improve the scalability and explainability of the PPO algorithm.