A new parameter identification method under non-white noise excitation using transformer encoder and long short-term memory networks (LSTMs) is proposed in the paper. In this work, the random decrement technique (RDT) processing of the data is equivalent to eliminating the noise of the raw data. In general, the addition of the gate in LSTM allows the network to selectively store data, which avoids gradient disappearance and gradient explosion to a certain extent. It is worthwhile mentioning that the encoder can learn the essence of data, which reduces the burden for the LSTM. More specifically, establish as simple LSTM structure as possible to learn the data of this essence to achieve the best training effect. Finally, the proposed method is used for simulation and experimental verification, and the results show that the method has the advantages of high recognition accuracy, strong anti-noise ability, and fast convergence rate. Specially, the results indicated appropriate accuracy proposed by deep learning combined with traditional method for parameter identification as well as proper performance of the proposed method.

Control method for the multi-axial half sine shock on random vibration test is studied in this article. From the Fourier spectrum analysis, it is found that the power of the half sine shock is mainly concentrated within a specific frequency band, which is only related to the sine frequency of the half sine shock. Compared with the broadband random signal, the frequency band occupies a small range. Therefore, a mixed signal separation method is proposed for the shock on random vibration control test. Firstly, the Fourier spectrum is divided into two parts, and then the Empirical Wavelet Transform (EWT) is used to extract the shock component from the mixed signal. Afterward, the random component is obtained by subtracting the shock component from the mixed signal. According to the deviations between each component and the specified references, the amplitudes of the shock component and the Power Spectral Densities (PSDs) of the random component are controlled by the amplitude correction algorithm and the matrix power control algorithm, respectively. An experiment with tri-axial shaker is implemented and the results show that the proposed method is feasible for the multi-axial half sine shock on broadband random vibration control test.

This paper presents a novel mechanical environmental test method that multi-output random strain responses are replicated instead of acceleration responses. In this work, the positive-definite strain reference spectral matrix and reference kurtoses for multiple control channels are defined first, and then three types of strain random signals including stationary Gaussian, stationary non-Gaussian, and non-stationary are generated. The inverse Fourier transform is used to obtain the stationary Gaussian strain random signals and the stationary non-Gaussian strain signals are obtained by the zero-memory nonlinear transformation. To synthetize non-stationary strain random signals, an amplitude modulation function is introduced with time-varying root mean square values. The correction algorithms are employed to update the drive signals for matching the strain responses compared with corresponding reference values. The simulation and experimental results prove the effectiveness and feasibility of the proposed strain response control procedure. The presented work provides new perspectives to investigate the fatigue behavior and life estimation of structure under controlled strain response environments for vibration fatigue tests.

To adaptively identify the modal parameters for time-invariant structures excited by non-white noise, this paper proposes a new operational modal analysis (OMA) method using hybrid neural networks. In this work, taking the acceleration response directly as the input data of the networks not only simplifies the data processing, but also retains all the characteristics of the data. The data processed by the output function is the output data of the network, and its peak corresponds to the modal frequency. The proposed output function greatly reduces the computational cost. In addition, a small sample dataset ensures that the hybrid neural networks identify the modal parameters with the highest accuracy in the shortest possible time. Interestingly, the hybrid neural networks combine the advantages of the convolutional neural network (CNN) and gate recurrent unit (GRU). To illustrate the advantages of the proposed method, the cantilever beam and the rudder surface excited by white and non-white noise are taken as examples for experimental verification. The results reveal that the proposed method has a strong anti-noise ability and high recognition accuracy, and is not limited by ambient excitation type.

Shock vibration environments frequently occur and threaten the reliability of products in aerospace engineering. This work deals with the shock response spectra replication control method for multi-shaker vibration test systems. Two synthetic methods for reconstitution of shock response signals with specified shock response spectra are presented. One is the random delay superposition method and the other is the filtering superposition method. The former utilizes a random time delay vector satisfying a normal distribution to simulate a shock response signal with symmetrical attenuated oscillation features while the latter is employed when the measured shock records are available. A time-domain inverse system method is formulated to obtain the multi-input drive signals based on a finite-difference equation. The multi-input multi-output frequency response functions are estimated first and then used to construct the inverse system. The multi-output shock response spectra are calculated by response signals from control sensors. A closed-loop iteration control procedure is provided to correct the drive signals according to the deviations between the responses and corresponding reference values. A numerical simulation and a triaxial vibration test are carried out and the results show the feasibility and effectiveness of the proposed methods.