To overcome the contradiction between the resolution and the measurement cost, various algorithms for reconstructing the sound field with sparse measurement have been developed. However, limited attention is paid to the computation efficiency. In this study, a fast sparse reconstruction method is proposed based on the Bayesian compressive sensing. First, the reconstruction problem is modeled by a sparse decomposition of the sound field via singular value decomposition. Then, the Bayesian compressive sensing is adapted to reconstruct the sound field with sparse measurement of sound pressure. Numerical results demonstrate that the proposed method is applicable to either the spatially sparse distributed sound sources or the spatially extended sound sources. And comparisons with other two sparse reconstruction methods show that the proposed one has the advantages in terms of reconstruction accuracy and computational efficiency. In addition, as it is developed in the framework of multitask compressive sensing, the method can use multiple snapshots to perform reconstruction, which greatly enhances the robustness to noise. The validity and the advantage of the proposed method are further proved by experimental results.
Driving mode switches of hybrid vehicles are significant events. Due to the different dynamic characteristics of the engine, motor, and wet clutch, it is difficult to coordinate torque fluctuations caused by mode switches. This paper focused on a control strategy for driving mode switches of plug-in hybrid electric vehicles (PHEVs) with a multi-disk wet clutch. First, the dynamic model of the PHEV was established, and a rule-based control strategy was proposed to divide the working mode regions and distribute the torque between engine and motor. Second, the dual fuzzy control strategy for a wet clutch and the coordinated torque control strategy for driving mode switches were proposed. The dual fuzzy logic control system consisted of the initial pulse-width modulation (PWM)’s duty cycle control and the changing rate of the PWM’s duty cycle control. Considering the difference in the dynamic characteristics between engine, motor, and wet clutch, a coordinated control strategy for the driving mode switches of PHEVs was put forward. Third, simulations of driving mode switches between pure electric driving mode and only engine driving mode were conducted. The results showed that the proposed control strategy could reduce the torque ripple and the jerk of the vehicle, completely satisfying the requirements of China. Finally, the control strategy for the motor-assisted engine starting process was tested on the bench. The experiment results indicated that the proposed control strategy was effective.
To solve the problem of sound field reconstruction with fewer measurement points, a sound field reconstruction method based on Bayesian compressive sensing is proposed. In this method, a sound field reconstruction model based on a combination of the equivalent source method and sparse Bayesian compressive sensing is established. The MacKay iteration of the relevant vector machine is used to infer the hyperparameters and estimate the maximum a posteriori probability of both the sound source strength and noise variance. The optimal solution for sparse coefficients with an equivalent sound source is determined to achieve the sparse reconstruction of the sound field. The numerical simulation results demonstrate that the proposed method has higher accuracy over the entire frequency range compared to the equivalent source method, indicating a better reconstruction performance and wider frequency applicability with undersampling. Moreover, in environments with low signal-to-noise ratios, the proposed method exhibits significantly lower reconstruction errors than the equivalent source method, indicating a superior anti-noise performance and greater robustness in sound field reconstruction. The experimental results further verify the superiority and reliability of the proposed method for sound field reconstruction with limited measurement points.
22 July 2022 in Computational Intelligence and Neuroscience
This paper solves the problem of difficulty in achieving satisfactory results with traditional methods of bearing fault diagnosis, which can effectively extract the fault information and improve the fault diagnosis accuracy. This paper proposes a novel artificial intelligence fault diagnosis method by integrating complementary ensemble empirical mode decomposition (CEEMD), energy entropy (EE), and probabilistic neural network (PNN) optimized by a sparrow search algorithm (SSA). The vibration signal of rolling bear was firstly decomposed by CEEMD into a set of intrinsic mode functions (IMFs) at different time scales. Then, the correlation coefficient was used as a selection criterion to determine the effective IMFs, and the signal features were extracted by EE as the input of the diagnosis model to suppress the influence of the redundant information and maximize the retention of the original signal features. Afterwards, SSA was used to optimize the smoothing factor parameter of PNN to reduce the influence of human factors on the neural network and improve the performance of the fault diagnosis model. Finally, the proposed CEEMD-EE-SSA-PNN method was verified and evaluated by experiments. The experimental results indicate that the presented method can accurately identify different fault states of rolling bearings and achieve better classification performance of fault states compared with other methods.