This study answers a practical question in renewable ventilation design: where does the exergy go when solar–geothermal preconditioning is combined with latent thermal buffering? A hybrid ventilation system coupling an earth–air heat exchanger (EAHE) with an air-based photovoltaic–thermal (PVT) unit is investigated, where phase change material (PCM) is embedded around the buried duct to act as a thermal buffer. Rather than reporting only annual energy savings, the work provides a risk-aware irreversibility breakdown that tracks exergy destruction across preconditioning, mixing, humidity control, and final conditioning. Several scenario analyses are conducted using hourly meteorological records collected over the 2019–2024 period. A bi-level optimization strategy is employed to simultaneously refine the system configuration and time-dependent operating conditions, including ventilation intensity and fresh-air intake ratio. To reduce occupant discomfort under critical conditions, a conditional value-at-risk criterion is incorporated to account for extreme thermo-hygrometric situations. The results indicate that PCM integration lowers system irreversibility. This behavior is mainly attributed to decreased ventilation demand and lower fresh-air intake. As a result, exergy destruction caused by air mixing and dehumidification is reduced. Although CVaR slightly increases during limited hot periods, overall discomfort risk remains about 50–60% below the baseline.
Given the vast array of food options available, food recommender systems play a crucial role in assisting users in discovering desired foods by analyzing their past preferences, represented as interactions between users and food items. To mitigate challenges stemming from data sparsity and cold start problems, these systems incorporate supplementary data resources in conjunction with rating data to enhance the accuracy of their recommendations. Furthermore, deep learning models offer a means to extract latent features from input data resources, thereby enhancing the accuracy of the recommendation process. Nevertheless, the performance of food recommender systems is intricately linked to the judicious selection and fusion of data resources. Previous research has overlooked this critical aspect in the design of food recommender systems. To address this challenge, this paper undertakes a comparative analysis of various data resources applicable to food recommender systems. We delve into the impact of leveraging deep neural networks to obtain deep representations of these data resources, as well as their fusion techniques, on the efficacy of food recommendations. Moreover, we comprehensively investigate the efficacy of various deep data representation fusion models in mitigating the challenges posed by data sparsity and cold start problems. To this end, we design extensive experiments utilizing two widely recognized food datasets. The findings indicate that the integration of additional data resources does not consistently improve recommendation accuracy. Also, varying combinations of these resources exert differential impacts on the performance of food recommender systems.
In transportation systems with mobile and modular facilities, how to make decisions related to facility location and capacity management during the planning horizon is critical; since the effective parameters in making these decisions change over time, it is required to adjust the locations and facilities’ operational capacity within the planning horizon to coordinate with the changing parameters. This paper presents a nonlinear programming formulation for a dynamic, modular, planar hub location problem when transportation demand is stochastic and its change in a continuous-time planning horizon is time-dependent. A dynamic programming-based hybrid solution method is presented to the proposed model, in which some polynomial-time algorithms (drop interchange, simple allocation, allocation improvement, Hyperboloid Approximation Procedure (HAP), and cost calculation) are used to calculate the cost of decisions in different states. The results of instances on the Australian Postal network (AP) dataset with up to a hundred nodes and eight hubs and sensitivity analysis are reported. The findings demonstrate that in the uncertain case, the total selected capacity for each hub in the entire planning horizon is more than in the case where the uncertain parameters take their mean value.
This study presented a procedure for predicting slope stability using four machine learning algorithms (extreme gradient boosting, support vector machine, logistic regression, and random forest). Based on a real-case database of 168 multinational slopes, this study analyzed the impact of six influential inputs: slope angle, friction angle, cohesion, height, pore pressure ratio, and unit weight. Notably, the results showed that the extreme gradient boosting algorithm surpassed other algorithms, achieving approximately 94
This paper proposes a comprehensive framework for cyber-resilient optimal energy management system (EMS) in smart grids. A complete 24-hour closed-loop operational cycle is modeled and simulated. It begins with data measurement via a real-time updated nonlinear Digital Twin, followed by transmission through SCADA/RTUs, detection of cyberattacks, accurate reconstruction of corrupted data, and finally EMS. The outputs of EMS are then sent back to the Digital Twin, which is dynamically updated to reflect the actual network conditions and generate accurate synthetic measurements for the next hour. This entire process is embedded within a 24-hour rolling optimization scheme. The EMS includes a power flow model integrated with various distributed energy resources (DERs), such as renewables, diesel generator, battery, electric vehicle, and controllable loads. It also incorporates and ensures all technical, security, and dynamic constraints of the grid and DERs. Unlike previous studies that focus only on isolated aspects such as attack detection, data estimation, or day-ahead energy management, this work implements the entire process in a unified and dynamic framework. The model functions effectively in networks where PMUs are unavailable, as is the case in most real-world distribution grids, because it is designed solely based on RTU and SCADA data. Detection and reconstruction of coordinated attacks rely on physics-based recalculation methods, utilizing grid topology and data from neighboring buses to improve accuracy. The proposed model is validated on the IEEE 33-bus test system, successfully detecting various attack scenarios targeting different parameters, locations, and times. It reconstructs the correct values with high precision and optimizes the network operation accordingly. This 24-hour rolling simulation demonstrates the practicality and robustness of the approach in enabling secure, cost-effective, and resilient energy management in modern smart grids.
