This work is concerned with the analysis of the effects of magnetic interactions beneath a vibrating beam excited by a rotating motor with unbalanced mass. The magnetic force arises from the interaction between a permanent magnet attached to the beam and an electromagnet supplied by a DC voltage source. The originality of the proposed configuration lies in using the magnitude and polarity of the electromagnet current to tune the magnetic interaction and, consequently, the nonlinear dynamical response of the beam in terms of vibration amplitude modification and displacement of the resonant frequency. After the modelling of the beam dynamics subjected to a rotating motor and a localized magnetic interaction force, the Galerkin-based modal reduction is used to derive a nonlinear equation governing the first vibration mode. The modal equation is solved numerically and its results compared to that of the direct numerical simulation of the nonlinear partial differential equation. Moreover, an experimental investigation is carried out on a bench structure with vibration amplitude and times series measured by an accelerometer. In the damped case, the magnetic current acts as an effective tuning parameter, shifting the frequency-response curve and modifying the vibration amplitude. The current polarity mainly determines the direction of the equilibrium displacement. For arbitrary mass parameters, a pronounced asymmetry with respect to the current sign is observed: positive currents lead predominantly to multi-periodic oscillations over the frequency range, whereas negative currents give rise to more complex dynamical regimes. For selected excitation frequencies, varying the magnetic current induces transitions between complex and periodic oscillatory regimes, indicating its potential for dynamic-state modification without necessarily implying vibration-amplitude suppression. In the absence of damping, the system exhibits richer dynamics for both parameter configurations, with extended regions of amplitude-modulated oscillations resulting from nonlinear interactions between forced and natural responses. Direct numerical simulations of the governing partial differential equation show good qualitative consistency with the reduced-order predictions, although quantitative discrepancies arise under some operating conditions. Experimental results reveal amplitude-modulated oscillations at intermediate excitation conditions, qualitatively consistent with the predicted dynamical trends. Quantitative analysis of the measured displacement signals yields vibration amplitudes ranging from 1.05 to 5.35 mm, which remain of the same order of magnitude as the numerical responses. Overall, the results show that the magnetic interaction provides both frequency detuning and dynamic-state modification, thereby offering a means of shaping the nonlinear response of beam-motor systems.
Blade failures cause much of the unplanned downtime of wind turbines, and most vibration-based methods assume a sensor on the damaged blade itself. This study tests whether the hub of a three-bladed rotor carries damage information between blades, so that a crack in one blade can be detected with sensors on the other two. Scaled NREL 5 MW blades were 3D-printed in PLA, mounted on a rigid hub and fitted with one accelerometer each. Transverse cracks were introduced at the root, transition zone and mid-span, and frequency response functions were measured by hammer-impact testing. The natural frequencies and damping ratios of modes 3, 4 and 6 were used as features for a multiclass support vector machine that identifies both the damage type and the damaged blade. A three-degree-of-freedom hub-coupled model supports the interpretation. The classifier reached an average accuracy of 99.62
This study evaluates the seismic performance of a five-story precast wall–slab–wall (WSW) structure and investigates whether precast wall–slab (WS) connections provide performance comparable to monolithic (ML) connections. The study also examines the applicability of pushover analysis (POA) for predicting seismic responses and comparing the seismic vulnerability of precast and ML structures. POA was performed on WSW structures incorporating two precast and two ML WS connection types. Three performance points (PPs), representing elastic, elastoplastic, and near-collapse conditions, were determined using the mean response spectrum derived from seven earthquake records representing far-field (FF), near-field with directivity effects (NFD), and near-field with fling-step effects (NFE). At each PP, the maximum base shear (BS), maximum top displacement (MD), and maximum inter-story drift ratio (MIDR) obtained from POA were compared with the mean responses from nonlinear time-history analysis (NLTHA). Additionally, fragility curves were developed for four damage states based on peak ground acceleration (PGA), following the HAZUS (2002) framework for nonlinear static analysis. Differences between POA and NLTHA responses reached up to 30
Panel flutter poses a critical safety challenge for supersonic aircraft. However, its behavior under the combined effects of random parametric error (RPE) and unknown external disturbance (UED) remains largely unexplored. This study numerically investigates the nonlinear dynamic behavior of a fluttering panel subjected to both RPE and UED, using the Galerkin method to discretize the governing equation into a high-dimensional system. The results establish a progressive need for high-dimensional modeling as system uncertainty increases. While a stable periodic vibration requires a three-order model, the introduction of a 42
Bearing vibration signals comprise random noise and intrinsic components, such as the rotational frequency and its harmonics. These components obscure fault signatures, thereby degrading the effectiveness of entropy-based feature extraction methods. It is challenging to distinguish fault-induced complexity variations from those caused by noise and intrinsic components. This work aims to develop an entropy-based feature extraction framework that suppresses these interferences and extracts discriminative fault features for reliable bearing fault diagnosis. A zero‑centered symbolic diversity entropy (ZSDE)-based feature extraction framework is proposed for bearing fault diagnosis. ZSDE integrates two stages of zero‑centering filtering (ZCF) and one stage of symbolic dynamic filtering (SDF). The first ZCF removes the rotational frequency and its harmonics from the synchronous vibration signal; the subsequent SDF suppresses random noise while preserving fault dynamics. Two phase spaces are then reconstructed using two different time delays: one based on the Poincaré section and the other on the conventional time‑delay criterion. For the second phase space, a second ZCF is applied to eliminate baseline drift in the embedding vectors. Finally, two complementary entropy measures (ZSDE‑1 and ZSDE‑2) are calculated and used as fault features. The proposed ZSDE is validated using simulated bearing signals and experimental data from a bearing test rig. Compared with nine existing entropy methods, ZSDE achieves relatively high diagnostic accuracy on both the simulated and experimental datasets. It also maintains stable diagnostic performance under different training sample sizes. Numerical experiments using a Duffing system further demonstrate favorable feature discriminability and competitive computational efficiency. The proposed ZSDE can effectively suppress the compound interference of random noise, the intrinsic components, and baseline drift by integrating a two‑stage ZCF and a one‑stage SDF. It provides a reliable and robust feature-extraction framework for bearing fault diagnosis and shows potential for extension to other dynamic systems.
