Construction sites are complex environments, and unsafe behaviors by workers, such as not wearing safety helmets or reflective vests, can easily lead to accidents. When using target detection technology to detect unsafe behaviors, the results are often unsatisfactory due to the complexity of the background and the small size of the targets. This paper proposes an unsafe behavior detection algorithm based on dual adaptive feature fusion. The algorithm is based on YOLOv5, introducing a front-end adaptive feature fusion module (FE-AFFM) at the head of the backbone network for deep data processing, improving the model’s feature extraction capability in complex backgrounds. Simultaneously, a back-end adaptive feature fusion module (BE-AFFM) is introduced at the tail of the network to strengthen feature fusion. In the experimental verification phase, this paper selects a self-made laboratory dataset and verifies the effectiveness of the improved algorithm through ablation experiments, algorithm comparisons, and heatmap analysis. The average accuracy of the improved algorithm is 3.6% higher than the baseline model, and the detection effect on small targets is significantly improved, meeting the actual needs of construction sites. This paper also selects the publicly available dataset SHWD for algorithm comparison experiments. The results show that the improved algorithm still has a significant advantage over mainstream algorithms, verifying the generalization ability of the improved model.

This paper studies resilient safe distributed optimization with practical prescribed-time performance for fractional-order multi-agent systems under false data injection attacks. Caputo dynamics are used to represent memory-dependent agent behavior. A local trimming mechanism is combined with gradient feedback, auxiliary coordination, a bounded time-varying gain, and a barrier-type safety correction. In contrast to an ideal singular prescribed-time design, the proposed bounded gain drives the normal-agent errors into an explicitly characterized neighborhood of the global optimizer at the assigned time, provided that a verifiable gain matrix is positive definite. Explicit residual bounds are derived in terms of the attack amplitudes, the trimming level, and the retained-neighbor redundancy. A fractional extremum argument is further used to establish forward invariance of the safe interval. The analysis proves boundedness of the physically relevant closed-loop signals, safety preservation, practical prescribed-time convergence, and robustness against bounded disturbances. Simulations on a sparse six-agent network demonstrate the roles of resilient information processing and barrier correction without requiring all-to-all communication.

This paper investigates connectivity-preserving formation control for high-order multi-agent systems under switching distance-dependent topologies, actuator faults and external disturbances. The admissible interaction edges are selected by a switching signal, while the actual edge weights vary continuously with inter-agent distances. To avoid unsafe jumps of algebraic connectivity at switching instants, a connectivity-certified switching condition is introduced. Between switching instants, an algebraic-connectivity-dependent barrier signal is incorporated into a recursive formation controller. A fixed-time extended state observer is designed to estimate the matched lumped uncertainty caused by actuator loss of effectiveness, additive faults and external disturbances. Explicit ultimate estimation-error bounds and a fixed convergence-time upper bound are derived. The effects of command filtering and smooth actuator saturation used in the numerical implementation are further represented by bounded implementation residuals in the stability analysis. Theoretical results establish boundedness of all closed-loop signals, uniform ultimate boundedness of formation errors and preservation of a prescribed connectivity margin under the stated hybrid safety conditions. Numerical simulations with five planar followers illustrate formation tracking, connectivity preservation, fault-tolerant disturbance rejection and observer estimation performance.

Monitoring the early-age evolution of cementitious materials is essential for ensuring the quality and reliability of concrete structures. However, most ultrasonic approaches rely on empirical correlations and lack a physics-based mechanism to describe the continuous viscoelastic transition during hydration. This study proposes a fractional-order ultrasonic sensing framework that couples a fractional Zener viscoelastic model with ultrasonic attenuation theory to quantitatively link microstructural evolution and measured acoustic responses. A custom ultrasonic measurement system was developed to capture real-time attenuation during hydration under different water-cement ratios. Results show that the fractional-order model achieves higher accuracy and robustness than classical integer-order and empirical models. The fractional parameter $\beta$ serves as a physically interpretable indicator that reflects the transition from viscous-dominated to elastic-dominated behavior and aligns with known hydration development. The proposed framework provides a compact, physics-informed sensing strategy for early-age characterization of cementitious materials and offers potential for intelligent construction and high-end structural monitoring.

Flexible manipulators combine low moving inertia with pronounced elastic vibration, and commonly used controllers often assume integer-order damping, direct velocity/full-state information, or plant-specific tuning without an explicit transient-error guarantee. This work considers a single-link manipulator with hereditary damping described by a physically parameterized fractional Kelvin–Voigt model. A model-assisted extended-state observer estimates the rigid-channel states and lumped uncertainty, while a modal differentiator reconstructs the flexible-mode velocity. The fractional nominal term is evaluated from measured modal history, and prescribed-performance tracking is combined with active modal damping under an explicit hub-torque constraint. A diffusive representation of the Caputo operator is used in the stability analysis, and the L1 discretization error is bounded explicitly. Under payload mismatch, external disturbances, and measurement noise, the controller achieves a tracking RMSE of $0.0059$ rad, an RMS tip deflection of $2.29$ mm, and $J_v=7.85\times {10}^{-5}$ m²s; the peak torque is $0.328$ N m under a $0.50$ N m limit. A conventional integer-order ADRC benchmark on the same plant gives $0.0518$ rad and $5.23$ mm, respectively. The prescribed tracking envelope is also maintained for an unloaded arm, a 300% payload increase, and an unmodeled second bending mode.