Search

Valve regulated lead-acid (VRLA) battery aging has threatened power supply reliability in cloud data centers, aimed at this problem, a method of an online battery state prediction method based on machine learning algorithms was proposed. The classic machine learning algorithm model was used to predict the battery status, and the prediction accuracy unextracted and the extracted feature values model was compared. The prediction accuracy of the model without extracted feature values ranges from 71.72% to 81.82%, and after extracting feature values, the prediction accuracy has improved by 10.10% to 20.20%, extracting feature values can improve prediction accuracy. The prediction model based on linear SVM method for extracting feature values is superior to other algorithms, with an accuracy of 96.46%. Experiment results show that machine learning algorithm based prediction models can be used for online VRLA battery state prediction in cloud data centers.

Summary: Weighted compact nonlinear schemes (WCNSs) are a popular family of high‐resolution shock‐capturing schemes for simulating compressible flows, of which the nonlinear interpolation procedure is dominant for the performance. In this work, a simplified weighting strategy is introduced for the nonlinear interpolation procedure. Firstly, an equivalent weighting formulation of WCNS is presented by explicitly including the whole‐point stencil into the set of candidate stencils. Secondly, motivated by the reorganization of WCNS, the WCNS‐CU6 scheme is achieved in a more straightforward way. Thirdly, by introducing a TENO selection procedure in the framework of WCNS‐CU6‐Simplified, a TCNS6‐Simplified scheme is proposed, the resolution of which is comparable with the excellent TENO6 scheme, while the computational cost is much lower. The simplified schemes exhibit more outstanding, at least comparable, fidelity than the original schemes, however, with superior characteristics in terms of efficiency and simplicity. A variety of benchmark test problems are studied to demonstrate the behaviour of the simplified weighting strategy. [ABSTRACT FROM AUTHOR]

Multiscale compressible turbulent flow is widely present in the aerospace field, and the co-existence of shock wave discontinuities and multi-scale turbulence in the flow field poses significant challenges for high-fidelity numerical simulation. To design high-order nonlinear scheme to improve the accuracy of complex turbulence simulation with stable shock wave capture, a weighted compact nonlinear scheme – targeted essentially non-oscillatory (WCNS-T) scheme is developed based on targeted essentially non-oscillatory scheme within the WCNS framework. Also, an improved variant of WCNS-T with smooth scale separation (denoted as WCNS-ST) is proposed from the perspective of both numerical convergence and accuracy. The numerical properties of different nonlinear weights are discussed in classical numerical problems, including spectral properties analysis, and two-dimensional Riemann problem. Then, the newly developed WCNS-ST scheme based on smooth scale separation was used to combine a unified turbulence model called Self-Adaptive Turbulence Eddy Simulation (SATES), which is applied to simulate subsonic flow past a 30P30N airfoil and supersonic base flow. The good convergence and numerical results are obtained using the newly developed WCNS-ST scheme, whereas WCNS-T shows worse convergence. In addition, adaptive dissipation control is further introduced for the WCNS-ST scheme, and the new variant (denoted as WCNS-STA) still shows good residual convergence and predicts more accurate results due to low numerical dissipation. The above results indicate that the proposed smooth scale separation and its WCNS-STA scheme have a good performance in SATES simulations of multiscale flow problems, making it more potential for complex engineering applications. [ABSTRACT FROM AUTHOR]

Based on one 5-point and two 2-point stencils, a nonlinear weighted scheme adaptive to the weighting of multi-order interpolation candidates for shock capturing is proposed. The JS weight function is adopted. The smoothness indicator of the 5-point stencil is specially designed with a higher-order leading term similar to the reference global smoothness indicator τ in the Z weighting. The design maintains that the nonlinear weights satisfy the sufficient condition for the scheme to avoid degradation at extreme points. The form of the weighting is also simplified, obviously reducing the weighting procedure's computational costs. Inviscid and viscous compressible flow problems are simulated to test the performance of the newly designed nonlinear weights based on the weighted compact nonlinear scheme. The numerical results show no obvious oscillations near the discontinuity, and the resolution of both discontinuities and smooth regions is better than JS, Z, and ZQ weighting. [ABSTRACT FROM AUTHOR]

Advanced array synthesis problems are usually nonconvex regarding the cost functions, which are rather difficult to solve. A numerical stochastic optimization approach based on the multiagent genetic algorithm (MAGA) is proposed for the synthesis of highly thinned arrays in this article. In particular, two modified techniques are introduced into the traditional MAGA, named as improved MAGA (I‐MAGA), to obtain the balance between exploration and exploitation ability. The proposed approach is utilized in some advanced antenna array synthesis problems, including the synthesis of sum and difference patterns and synthesis of several highly thinned planar arrays with different apertures. Numerical results indicate that the proposed I‐MAGA has an attractive exploration and exploitation ability in the synthesis of highly thinned arrays. Besides, owing to the independent multiagent system, the I‐MAGA is also able to obtain better solutions with a much smaller population size and less computational cost. [ABSTRACT FROM AUTHOR]