As power systems decarbonise and digitalise, electric power marketing (EPM) is becoming a key interface between regulation, system operation and sustainable-energy deployment. However, many utilities still rely on human agents and rule-based chatbots with fragmented knowledge bases, which struggle with long, cross-scenario dialogues and often fail to provide compliant and verifiable responses. Although large language models (LLMs) show strong conversational ability, existing benchmarks underrepresent the sector-specific terminology, regulatory reasoning and multi-turn stability required in EPM. To address this gap, we present ElectriQ, a large-scale benchmark for LLM evaluation in EPM. ElectriQ contains over 550k multi-turn dialogues across six service domains and 24 sub-scenarios, and defines a unified evaluation protocol combining human ratings, automatic metrics, and two dedicated stress tests: Statutory Citation Correctness (SCC) and Long-Dialogue Consistency (LDC). For reproducible comparison, we further implement SEEK-RAG as a reference baseline to support evidence-grounded responses on ElectriQ. Experiments on 13 LLMs show that domain-aligned 7B models can match or surpass much larger models at lower computational cost, providing an auditable basis for deploying LLM-based EPM assistants in demand-side management, renewable integration and resilient grid operation.
Blade icing can reduce wind-energy capture efficiency and increase the structural and operational risks of wind turbines. This study proposes an acoustic-signal-based method and prototype system for the online assessment of wind turbine blade icing potential. The objective is to estimate, at each monitoring time, the probability that a blade is under a high-risk icing condition using blade-passage acoustic signals together with environmental and operational parameters. The task considered in this study is therefore online icing-potential assessment rather than diagnosis of general blade faults or long-horizon forecasting of future ice accumulation. The model outputs an estimated icing-potential probability, which is converted into a high-risk icing label through probability-threshold and temporal-window processing. Prototype testing provides preliminary evidence of the feasibility of the proposed method. The study provides a technical reference for online blade-icing monitoring and high-risk icing warning under extreme weather conditions.
The petroliferous Aryskum Depression, located within the South Torgai Basin, Kazakhstan, comprises several grabens up to 6 km deep, separated by horsts with more than 4 km of structural relief. Spatial analysis of reported reservoir and measured bottomhole temperatures from oil and gas wells revealed elevated geothermal gradients, ranging from 29 to 45 °C/km. An underlying active mantle plume is the main source of this anomalous heat flux. Higher gradients (>35 °C/km) are likely related to additional factors, including radiogenic heat production from granites in the basement, high thermal conductivity of the crust (including intrusive bodies), and convective heat transfer by hydrothermal fluids circulating through open faults within the highly fractured basement and sedimentary basin-fill. The overlay of basement lithotypes, faults, and reservoir-quality sands allowed for robust delineation of isotherms and prediction of geothermal reservoirs. Favorable conditions exist across the basin for both direct electricity generation with temperatures of ∼150 °C at depths of ∼4.3 km, and direct heat utilization from shallower aquifers. The study demonstrates how integration of data from various sources, in a structural and sedimentological framework, can advance exploration for blind geothermal systems in analogous basins worldwide.
This paper presents a robust multi-objective optimization framework based on the Slime Mould Algorithm (SMA) for a stand-alone PV–Battery–Electrolyzer system targeting green hydrogen production. Utilizing real-world solar generation profiles from the Bursa Technical University Smart Grid Laboratory, the study simultaneously maximizes green hydrogen yield and minimizes energy waste under stochastic weather conditions across six distinct meteorological scenarios encompassing summer and winter seasons under sunny, cloudy, and rainy conditions. The proposed SMA-based energy management framework achieves a peak daily hydrogen production of 58.84 kg with an average system efficiency of 94.28%, while maintaining near-perfect solar utilization even during highly volatile periods. A single-algorithm sensitivity analysis identifies a 500 kWh BESS capacity as the optimal trade-off point, delivering the highest hydrogen yield (49.8 kg/day) while minimizing energy curtailment. Furthermore, a rigorous benchmarking study against Genetic Algorithm (GA) and Particle Swarm Optimization (PSO) under identical conditions confirms the SMA’s holistic superiority, achieving the highest mean hydrogen production (36,012.2 g), the lowest mean energy waste (53.3 kWh — approximately 3.2× lower than GA and 8.1× lower than PSO), and the highest mean solar utilization (97.7%) simultaneously across all scenarios. A two-dimensional sensitivity analysis further demonstrates that the SMA is the only algorithm to scale reliably and monotonically in both hydrogen yield and solar utilization as BESS capacity and peak solar generation increase, confirming its superior robustness and scalability for future system expansions.
Achieving climate neutrality in the building sector necessitates a well-balanced mix, entailing measures that reduce the space heating demand and measures to decarbonize the heat supply system. In non-residential buildings, the space heating demand can be reduced through both renovation measures targeting the building envelope and heat recovery from the ventilation system. In order to avoid CO2 emissions, the implementation of the entire set of retrofit measures should be aimed at, but is often not practically feasible due to financial and other constraints. This issue may be addressed with a hybrid heat pump system, wherein the existing, fossil-fueled boiler continues to be operated to cover the residual thermal load until other retrofit measures have reduced the space heating demand to a level that it can only be covered by the heat pump. The proposed approach combines a techno-economic assessment of retrofit measures with a detailed simulation of the hybrid heat pump system across different stages of implemented retrofit measures. The process of bundling and sequencing gives rise to a multitude of combinatorial permutations of possible retrofit roadmaps. To apply the methodology to a real building as a case study, realistic requirements and constraints are transferred into filters to obtain a manageable number of possible roadmaps as a basis for exemplary roadmaps as scenarios. Hence, this approach is suitable for a techno-economic assessment of adequate measure bundles that ensure the transformation to carbon-neutral buildings despite limited resources.
