Variable vibration is difficult to control due to the uncertainty and variability of vibration frequency. Vibration absorbers are widely used in engineering, however, the efficiency of mathematical design method decreases significantly once the external vibration changes. In order to suppress the variable frequency vibration, a novel self-learning tuning method for the vibration absorbers is proposed based on a large number of experimental vibration data, which contains the actual information of the optimal absorber parameters. The self-learning tuning method proposed can identify the external excitation frequency and label the optimal stiffness of the absorbers adaptively. In addition, model training can help achieve optimum vibration suppression through reasonable stiffness tuning for different variable low frequency excitation. Two vibration absorbers are designed with negative electromagnetic stiffness, and the effective frequency band can be decreased by applying current into the coil automatically. The vibration absorbers are slightly askew installed on the primary system to achieve bidirectional vibration attenuation. The experimental results indicate that the maximum vibration attenuation ratio is improved from 87.96% to 95.88% along Z-axis and 79.21% to 89.93% along X-axis in the bandwidth of 7–9 Hz.

由于振动频率的不确定性和多变性，变振动难以控制。减振器在工程中应用广泛，但是一旦外部振动发生变化，数学设计方法的效率就会显着降低。为了抑制变频振动，基于大量实验振动数据，提出了一种新的减振器自学习整定方法，其中包含最佳减振器参数的实际信息。所提出的自学习调谐方法可以识别外部激励频率并自适应地标记吸收器的最佳刚度。此外，模型训练可以通过针对不同的可变低频激励进行合理的刚度调整，帮助实现最佳的振动抑制。两个减振器采用负电磁刚度设计，可通过自动向线圈通电来降低有效频带。减振器微斜安装在一次系统上，实现双向减振。实验结果表明，在 7-9 Hz 的带宽内，最大振动衰减率沿 Z 轴从 87.96% 提高到 95.88%，沿 X 轴从 79.21% 提高到 89.93%。