Abstract

Timely and accurate building extraction from disaster imagery is important for emergency response, damage assessment, and post-disaster recovery. Staring satellite video provides stage-specific observations of the same affected area as a disaster evolves. However, building appearances vary substantially across stages, from intact structures to deformed buildings and fragmented remnants, resulting in pronounced stage-dependent within-class variation. Existing methods are commonly developed using single-date imagery, and their reliability across heterogeneous damage conditions remains insufficiently investigated. To address this problem, we propose PGDGNet, a Prior-Guided Detail–Global Fusion Network for frame-wise multi-stage building extraction. PGDGNet employs a dualbranch CNN–Transformer architecture to capture local structural details and global semantic context. Its Detail–Global Feature Fusion (DGFF) module applies branch-specific attention and spatial gating to reconcile original and attention-enhanced features before local–global integration. A joint color-and-boundary prior-guidance strategy provides complementary appearance and contour cues when structural information is weakened, while boundary-aware supervision facilitates the delineation of incomplete buildings. A patch-level dataset is constructed from frames representing the early, intermediate, and late stages of the 2018 Indonesian tsunami–earthquake event. Experiments show that PGDGNet achieves the highest Building IoU and recall across all three stages. Compared with the strongest baseline, CFFormer, PGDGNet improves Building IoU by 0.59, 0.63, and 0.81 percentage points and recall by 1.45, 0.80, and 1.08 percentage points in the early, intermediate, and late stages, respectively. The results demonstrate robust and reliable performance across disaster stages.

Abstract

Ultra-high-performance concrete (UHPC) offers superior mechanical performance but remains limited by high cement content, sustainability concerns, and the lack of reliable constitutive models capable of capturing its full stress–strain response including pre-peak and post-peak behaviour. This study presents an integrated experimental and constitutive modelling investigation of UHPC incorporating ultra-fine waste glass powder, micro silica, fly ash, and different fibre reinforcement. Two UHPC mix designs with a water-to-cementitious material ratio of 0.20 were developed, yielding 24 independent mixtures. Cement was partially replaced with 20% ultra-fine waste glass powder and combined with either micro silica or fly ash. Fibre reinforcement was introduced independently using 2% straight steel fibres and 1.5% crimped polypropylene fibres to evaluate flowability, viscosity, uniaxial compressive strength, modulus of elasticity, splitting tensile strength, flexural strength and drying shrinkage behaviour. Experimental results demonstrate that UHPC with micro silica, UFWGP and steel fibres, prepared following the ACI 239 R-18 mix design, achieved the highest 28 days compressive, splitting tensile and flexural strength of 144.8, 12.8 and 36.1 MPa, respectively. Steel fibre-reinforced mixes exhibited up to a 42% increase in compressive strength and significantly enhanced post-peak ductility compared to the control mix (without fibre). Crimped Polypropylene fibre-reinforced UHPC attained compressive strength up to 129.7 MPa, provided improved strain capacity, and effectively reduced drying shrinkage, highlighting their suitability for crack control and deformation mitigation. The incorporation of UFWGP consistently improved flowability, reduced viscosity, and enhanced mechanical performance, while fly ash improved rheological behaviour but resulted in lower earlyage strength compared to micro silica-based mixes. To capture the compressive response of UHPC, an elasto-damage model previously proposed by Khan and Zahra was reformulated by recalibrating the compression damage parameter (β) using experimentally derived and literature based compressive strength and elastic modulus. The proposed model reproduces both pre-peak and post-peak stress–strain behaviour, with prediction errors generally within ±7% of experimental results ranging from 72.4 to 148.5 MPa. The findings provide robust experimental evidence and a validated constitutive framework for the sustainable design and structural application of UHPC

Abstract

The consensus control problem to be discussed in this paper is concerned with nonlinear multi-agent systems with unknown dynamics, unknown fractions, time-varying delays, and input saturation limits. In order to address these difficulties, an adaptive control system is established, incorporating a recurrent general type-2 fuzzy neural network (RGT2FNN) with a biogeography-based optimization (BBO) algorithm. The RGT2FNN is used to model nonlinear functions that are not known offline, and the BBO algorithm also optimizes the parameters of the fuzzy network and performs offline identification of the fractional order by minimizing multi-step model prediction error. In order to make the model more resistant to modeling uncertainties, time-varying delays, and actuator saturation effects, a LMI-based compensator is proposed to ensure the stability of the closed-loop. Lyapunov analysis guarantees the boundedness of consensus errors. The simulation findings prove that the suggested methodology can attain an accurate consensus tracking and strong performance when the uncertainties are harsh, the delays may change with time, and the limits of input saturation are taken into account.OPEN ACCESS Received: 06/01/2026 Accepted: 16/03/2026