Model-based predictive control provides an effective approach for improving building energy efficiency by optimizing heating, ventilation, and air-conditioning operation while maintaining indoor thermal conditions. However, its performance strongly depends on the availability of an accurate and generalizable prediction model of building thermal dynamics. Existing studies have focused on individual buildings, where each building is modeled and optimized independently. Recent studies have demonstrated the potential of federated learning (FL) for a range of building applications. However, most existing approaches remain predominantly data driven. This study proposes a federated physics-informed neural network for generalized indoor temperature prediction across multiple building zones. The global RC model was developed and then incorporated into the neural-network loss function as a physics-based constraint, while FL was employed to develop a shared global prediction model without pooling the raw datasets of individual clients. The developed global model demonstrated good single-step-ahead prediction performance across the four target zones, with root-mean squared error (RMSE) values ranging from 0.36 °C to 0.59 °C, mean absolute error values from 0.25 °C to 0.57 °C. For multi-step-ahead prediction, the average RMSE was approximately 0.68 °C during the estimation period and 0.99 °C during validation. Moreover, when the zone excluded from the federated training process and evaluated as an unseen client the proposed model achieved RMSE values of 0.74 °C and 0.86 °C for one-step-ahead and multi-step-ahead predictions, respectively, compared with 0.86 °C and 0.98 °C for the conventional data-driven FL model.
This study examines students’ preferences for sensory elements in designing a dedicated restorative space within a university environment. The research employs a sensory-scale Restorative Affordances framework, in which the five affordances of Retreat, Engage Softly, Spatial Coherence, Express Identity, and Tailored Fit were derived from the integration of Attention Restoration Theory (ART), Theory of Supportive Design (TSD), and Perceived Sensory Dimensions (PSD). Data was collected from 30 interior design students through open-ended questionnaires and analyzed using a combined frequency-based method with Voyant Tools and manual qualitative analysis. Eleven sensory themes were identified and grouped into Visual elements, including lighting, color, pattern, and view; Ambient elements, including sound, air, thermal condition, and scent; and Haptic elements, including material, texture, and tactility. Visual elements were the most frequently mentioned, followed by ambient elements, while haptic elements were less emphasized. Lighting (24.50
Construction claims in engineering, procurement, and construction projects are often presented as monetary totals, while the records needed to interpret their scope, valuation, and time consequences are distributed across different document types. This study examined three claim bundles from the coal-handling works of a thermal power plant in the Republic of Korea: an additional sump-pit structure, revised shed paving quantities, and a disputed work stoppage. A documentary evidence-chain method separates the initiating event, version of the work and quantity, pricing basis, parties’ positions, and schedule evidence. Cost figures are expressed as percentages of a specified subtotal within each bundle. Materials account for 56.70
Radio utterances during concrete pouring convey work progress, equipment conditions, hazard signs, and control actions in short units that may include omitted subjects and site expressions. This study presents a proof of concept for a two-layer ontology framework that structures these utterances into linked semantic and situational records. Starting from timestamped radio text linked to terminal and assigned-role information, the first layer classifies expressions into seven semantic axes: work, act, object, target, phenomenon, interjection, and profanity. The second layer derives hazard signals, control actions, and situation states using semantic relations, action-expression conditions, and chronological context. The procedure is implemented through lexical dictionaries and rules and demonstrated with a researcher-constructed one-day virtual log of 25 utterances. Worked examples show how object and target distinctions lead to different hazard or control-action interpretations, and how subsequent reports and instructions are linked to pressure resolution and restart restrictions. Reference comparisons and error analysis examine internal consistency and unresolved mappings within this development case. An exploratory LLM comparison illustrates challenges in mapping field expressions to the defined semantic axes. The proof of concept demonstrates a procedure for converting linked radio text into structured records that can support later review of communicated conditions and instructions. The contribution is the explicit representation and implemented processing sequence, with quantitative metrics serving as diagnostic evidence for the case. Audio acquisition, speech recognition, and field performance remain subjects for subsequent validation.
This study investigated composite formulation design strategies to address the decline in mechanical properties that occurs during the process of ensuring the fire-resistance performance of two-component polysulfide sealants. To overcome the reduction in matrix cohesion and adhesion caused by the high loading of inorganic flame retardants, amino-silane-based adhesion promoters and high-aspect-ratio cotton fiber powder were introduced, and the resulting thermal and mechanical reinforcement mechanisms were evaluated. The experimental results confirmed that, owing to the formation of chemical interfacial cross-linking and three-dimensional stress transfer pathways, the tensile strength significantly increased to 37 kgf/㎠ (3.63 MPa) (23.3
