Active vibration reduction technology for mixed spectrum noise

In engineering systems, vibration and noise arise from the coupling of mechanical energy across multiphysical fields. These phenomena are governed by complex excitation sources, nonlinear transmission paths, and system-specific modal characteristics. Consequently, purely narrowband (line-spectrum) or purely broadband (continuous-spectrum) vibration and noise signals are rare under operational conditions. Instead, structural resonances and other dynamic effects produce a mixed spectrum, characterized by narrowband line-spectrum components superimposed on a broadband continuous-spectrum background. This mixed-spectrum behavior necessitates advanced control strategies to effectively mitigate vibration and noise. Previous studies on active noise control technology mainly focused on the control of narrowband vibration and energy-eminent line spectrum noise with periodic characteristics, while they paid less attention to broadband vibration noise (with nonperiodic time-varying characteristics) that reduces the overall vibration damping effect. Vibration control for actual working conditions must break through the unimodal thinking of traditional algorithms and establish broadband and narrowband synergistic control algorithms so as to improve the overall vibration damping effect of the equipment.

Aiming at the actual vibration conditions close to the wideband and narrowband hybrid vibration model, a vibration reduction device based on the pretrained selection coefficient model of the mixed-spectrum hybrid vibration noise active control (MSN-HVNC) algorithm is designed and successfully used in the vibration abatement active control experiments. The hybrid control algorithm is used for vibration noise abatement. Moreover, the wideband noise control subsystem uses a pretrained neural network model to select filter coefficients for updating the coefficients of the filtered-x least mean square (FxLMS) algorithm and thereby controlling the wideband noise, while the narrowband noise control subsystem abates the line-spectrum noise, which is concentrated in energy. The overall damping level of the algorithm is measured in terms of the residual vibration noise to update the controller weights.

The damping effect of the vibration-damping device based on the MSN-HVNC algorithm is 23.0 and 21.3 dB under single-frequency 50 Hz excitation in Case 1 and single-frequency 75 Hz excitation in Case 2, respectively, and the damping effect under mixed excitation vibration signals in multisource coupled vibration scenarios in Case 3 is 12.0 dB. The average damping effect of the MSN-HVNC algorithm is better than that of the FxLMS algorithm for both the single-frequency narrowband line-spectrum noise case and the complex vibration noise case. The MSN-HVNC algorithm is better than the FxLMS algorithm for both single-frequency narrowband linear spectrum noise and complex vibration noise conditions and exhibits a good damping effect for vibration noise. The vibration damping device accelerates the speed of noise control in the form of integrated control with the pretrained coefficient model and independent subsystems, which better meets the engineering needs of modern equipment intelligence and high efficiency, and provides an innovative solution for vibration and noise control in fields such as ship power systems.