Neil Yorke-Smith

Road freight is a major source of transport-related greenhouse gas emissions, making heavy-duty vehicle electrification important for reducing direct CO$_2$ emissions. However, highway HDEV charging concentrates large energy demands into short charging windows, resulting in megawatt-scale power requirements that can exceed available grid capacity. Effective operation therefore requires an energy management system (EMS) that coordinates charging demand with local generation, battery energy storage system (BESS) flexibility, and time-varying grid availability. Optimization-based EMSs can explicitly enforce constraints, but repeated online optimization increases computational effort as problem size and prediction horizon grow, while nonlinear component models often require tractable approximations. Reinforcement learning (RL) offers an alternative by learning the control policy offline and generating actions with low online computational effort. However, conventional RL does not inherently enforce physical feasibility; operational limits are typically handled through reward penalties or downstream action correction, which can leave the policy dependent on external feasibility handling or distort competing objectives when large penalties are required. To address this limitation, this study develops a physics-informed proximal policy optimization (PI-PPO) EMS for a highway HDEV charging station with PV, BESS, and a time-varying grid connection. Physics-based residuals associated with grid and BESS constraints are embedded directly in the actor optimization, while forecasts of HDEV demand, PV generation, electricity prices, and grid availability support anticipative control. PI-PPO is evaluated against conventional PPO, single step MILP, and rolling-horizon MILP. The results show that PI-PPO achieves a better balance between charging service and physical feasibility, with markedly improved grid compliance and BESS-feasible actions. It captures part of the anticipative and constraint-aware behaviour of rolling-horizon MILP while retaining fast online inference. Moreover, the learning-based formulation directly incorporates nonlinear BESS ageing, avoiding the degradation underestimation observed with the MILP-compatible approximation.

The electrification of on-road fleet logistics promises improved air quality, lower noise emissions, major climate benefits, increased energy flexibility through the use of locally generated electricity and reduced dependence on imported fuels. However, batteryelectric vehicles can introduce operational planning challenges not present with internal combustion engine vehicles, including heterogeneous charging speeds, exposure to volatile electricity prices, and scarcity in infrastructure. Managing these complexities requires solutions that balance cost efficiency and robustness, supported by sector coupling between transport and electricity systems. This paper reviews the current state of digital systems for operational decision-making in electric fleet management througha grey literature analysis, drawing on practitioner-oriented sources such as industry reports, company documentation, and techni-cal blogs that reflect real-world practices and developments. We identify key trends and gaps, providing insights to guide futureresearch and development.

Deep learning-based crack segmentation is often limited by scarce annotated data and the limited controllability of existing augmentation methods. To address these challenges, this study proposes CPT-Seg, a unified framework that integrates controllable crack pattern transfer with downstream crack segmentation. A crack pattern transfer module is developed to decouple crack geometry from surface texture through mask-guided fusion, Stable Diffusion Inpainting, and LoRA fine-tuning, enabling realistic crack generation on diverse surfaces while preserving geometric fidelity. The generated images are incorporated into a two-stage transfer learning strategy, where synthetic crack images are used for pre-training and real annotated images for fine-tuning. Experimental results demonstrate that the proposed crack pattern transfer approach achieves a PSNR of 18.22 dB, SSIM of 0.550, LPIPS of 0.351, and FID of 134.74, demonstrating strong image quality and structural consistency. When applied to crack segmentation, the data generated improved the IoU from 88.0% to 88.9% and the F1-score from 92.5% to 93.5% compared with training on real data alone. Cross-dataset validation further confirms robust generalization across concrete and pavement crack datasets. The proposed framework provides an effective solution to alleviate data scarcity and dataset bias, bridging generative AI and vision-based structural health monitoring.

Simulation–optimization is often used in enterprise decision-making processes, both operational and tactical. This paper proposes a data-driven, declarative approach to enterprise optimization. In our generic simulation–optimization approach, a declarative constraint satisfaction problem (CSP) is automatically customized given a simulation model and a problem instance. We construct the declarative model by training a neural network on the simulation model and embedding the trained network in the CSP. The approach allows the embedding of a wide variety of complex problems, is able to handle multi-objective problems, is flexible to changing multiple objectives simultaneously, and allows the modeler to focus on what problem needs to be solved by a computer instead of how the computer should solve it. We demonstrate the value of this neuro-symbolic approach in experiments on two problem domains. Furthermore, we show how low discrepancy sampling, weighted composite objectives, and linear activation functions can improve the performance of neural embeddings in CSP.

Uncertainty quantification in reinforcement learning can greatly improve exploration and robustness. Approximate Bayesian approaches have recently been popularized to quantify uncertainty in model-free algorithms. However, so far the focus has been on improving the accuracy of the posterior approximation, instead of studying the accuracy of the prior and likelihood assumptions underlying the posterior. In this work, we demonstrate that there is a cold posterior effect in Bayesian deep Q-learning, where contrary to theory, performance increases when reducing the temperature of the posterior. To identify and overcome likely causes, we challenge common assumptions made on the likelihood and priors in Bayesian model-free algorithms. We empirically study prior distributions and show through statistical tests that the common Gaussian likelihood assumption is frequently violated. We argue that developing more suitable likelihoods and priors should be a key focus in future Bayesian reinforcement learning research and we offer simple, implementable solutions for better priors in deep Q-learning that lead to more performant Bayesian algorithms.

Training neural networks (NNs) using combinatorial optimization solvers has gained attention in recent years. In low-data settings, the use of state-of-the-art mixed integer linear programming solvers, for instance, has the potential to exactly train an NN while avoiding computing-intensive training and hyperparameter tuning and simultaneously training and sparsifying the network. We study the case of few-bit discrete-valued neural networks, both binarized neural networks (BNNs) whose values are restricted to ±1 and integer-valued neural networks (INNs) whose values lie in the range [ ]. Few-bit NNs receive increasing recognition because of their lightweight architecture and ability to run on low-power devices: for example, being implemented using Boolean operations. This paper proposes new methods to improve the training of BNNs and INNs. Our contribution is a multiobjective ensemble approach based on training a single NN for each possible pair of classes and applying a majority voting scheme to predict the final output. Our approach results in the training of robust sparsified networks whose output is not affected by small perturbations on the input and whose number of active weights is as small as possible. We empirically compare this BeMi approach with the current state of the art in solver-based NN training and with traditional gradient-based training, focusing on BNN learning in few-shot contexts. We compare the benefits and drawbacks of INNs versus BNNs, bringing new light to the distribution of weights over the [ ] interval. Finally, we compare multiobjective versus single-objective training of INNs, showing that robustness and network simplicity can be acquired simultaneously, thus obtaining better test performances. Although the previous state-of-the-art approaches achieve an average accuracy of [ ] on the Modified National Institute of Standards and Technology data set, the BeMi ensemble approach achieves an average accuracy of 68.4% when trained with 10 images per class and 81.8% when trained with 40 images per class while having up to 75.3% NN links removed. History: Accepted by Andrea Lodi, Area Editor for Design & Analysis of Algorithms—Discrete. Funding: This research was partially supported by the European Union Horizon 2020 Research and Innovation Programme [Grant 952215]. The work of A. M. Bernardelli is supported by a PhD scholarship funded under the “Programma Operativo Nazionale Ricerca e Innovazione” 2014–2020. Supplemental Material: The software that supports the findings of this study is available within the paper as well as from the IJOC GitHub software repository ( https://github.com/INFORMSJoC/2023.0281 ). The complete IJOC Software and Data Repository is available at https://informsjoc.github.io/ .