一种基于粗糙K-均值的椭球基函数神经网络

改进了一种椭球基函数神经网络,它与经典椭球单元神经网络的结构不同,而与径向基函数神经网络结构类似,即它有一个隐含层,并且隐层单元采用椭球基函数,区别于RBF网络的高斯函数。本文采用粗糙K-均值方法求取椭球函数的中心,并给出了该方法中确定初始阈值的步骤。这种改进方法不但使对输入空间的划分局部作用,而且划分区域封闭有界。因此,改进的神经网络具有较好的函数逼近能力和模式识别能力。仿真实验验证了该椭球基函数神经网络的正确性和有效性。     

正态-复合非时齐Poisson过程模型结构可靠性当量正态设计

讨论强度R～N（μ_R，_R），应力｛S（t），t∈［0，T］｝为非时齐Poisson过程模型结构可靠性当量正态设计。获得应力在设计基准期［0，T］内的最大值概率分布及统计参数估计；正态－复合非时齐Poisson过程模型结构可靠度及结构可靠性当量正态设计表达式。     

一种研究圆柱度目标函数单谷性的新方法

提出了一种研究圆柱目标函数单谷性的新方法，圆柱度目标函数单谷性的问题目前研究得还少，而弄清楚圆柱度目标函数是否是单谷函数对于用优化原理正确地计算其极值是非常重要的。     

化学反应过程方式的判据及反应方向的判据

本文对化学反应过程的方式及方向判据问题进行了简要分析,并指出用△G~Q代替△G判断化学反应过程的粗略界限。     

J—8ACT验证机平尾助力系统抖振分析

本文对J-8ACT验证机纵向数字电传操纵系统所使用的平尾助力系统的抖振原因和抑制措施进行了线性和非线性分析。证明了引起助力系统抖振的主要原因,除由流量特性决定的过大的流量系数外,对于内反馈式的助力系统来说,不可避免的安装支承间隙是一个重要因素。机械输入式液压伺服裝置的一些重要特性是由滑阀摩擦引起的。这种受机械反馈影响的特殊摩擦力,具有增加液压作动器的增益和滞后作用。由于这个特点,它能抑制/消除助力系统抖振。     

Superclustering and the large-scale structure of the universe

The spatial distribution of a complete redshift sample of nearby rich clusters of galaxies is studied. Strong superclustering, extending to very large scales (100h⁻¹Mpc), is observed in the cluster distribution. In particular, we determined and discuss below the following studies:I. The spatial correlation function of rich clusters of galaxies.II. The construction of a complete catalog of superclusters and a study of their growth (percolation).III. The finding of large-scale superclusters surrounding the giant galaxy void in Bootes.IV. The finding of a 300 Mpc void of rich clusters of galaxies.All the above studies provide evidence for the existence of large-scale structure in the universe, and reveal some of its characteristics. These findings are of importance to models of the formation of galaxies and structure in the universe.     

高效空间应用燃料电池热能利用技术研究

首先介绍了空间应用燃料电池高效散热及温控技术。高精度控温主要是利用被动散热技术,将产生的热量快速传导至外部冷却介质,从而保证产热面的温度均匀性,实现电池稳定工作。通过嵌入式被动散热和双端散热的冷却方式,实现电堆内部最大温差小于5℃,换热系数达1000W/(m·K)。     

Analysis on Output Power for Multi-direction Piezoelectric Vibration Energy Harvester

To predict the performance of multi-direction piezoelectric vibration energy harvester,an equation for calculating its output power is obtained based on elastic mechanics theory and piezoelectricity theory.Experiments are performed to verify theoretical analysis.When the excitation direction is along Y direction,a maximal output power about 0.139 mW can be harvested at a resistive load of 65kΩ and an excitation frequency of 136 Hz.Theoretical analysis agrees well with experimental results.Furthermore,the performance of multi-direction vibration energy harvester is experimentally tested.The results show that the multi-direction vibration energy harvester can harvest perfect energy as the excitation direction changes in XY plane,YZ plane,XZ plane and body diagonal plane of the harvester.     

Energy principle and material removal sequence optimization method in machining of aircraft monolithic parts

In the machining process of aircraft monolithic parts, the initial residual stress redistribution and structural stiffness evolution often lead to unexpected distortions. On the other hand, the stress redistribution and stiffness reduction during the machining process depend on the material removal sequence. The essence of the stress redistribution is releasing the initial elastic strain energy. In the present study, the influence of the material removal sequence on the energy release is studied. Moreover, a novel optimization method is proposed for the material removal sequence. In order to evaluate the performance of the proposed method, the mechanism of the machining distortion is firstly analyzed based on the energy principle. Then a calculative model for the machining distortion of long beam parts is established accordingly. Moreover, an energy parameter related to the bending distortion and the procedure of the material removal sequence optimization is defined. Finally, the bending distortion analysis and material removal sequence optimization are performed on a long beam with a Z-shaped cross-section. Furthermore, simulation and experiments are carried out. The obtained results indicate that the optimized sequence results in a low distortion fluctuation and decreases the bending distortion.     

Multiscale topology optimization using feature-driven method

This paper presents a multiscale design method for simultaneous topology optimization of both macrostructures and microstructures. Geometric features are extended as design primitives at both macro and micro scales and represented by Level Set Functions (LSFs). Parameters related to the locations, sizes, and orientations of macro and micro features are considered as design variables and optimized simultaneously. In the overlapping areas of different macro features, embedded microstructures are optimally figured out as the solution of the corresponding sub-optimization problem. In this study, the eXtended Finite Element Method (XFEM) is implemented for structural and sensitivity analyses with respect to design variables. This method has the advantage of using a fixed grid independent of the topology optimization process. The homogenization procedure is applied to calculate the effective properties of considered microstructures in each macro feature. Numerical examples are presented to illustrate the effectiveness of the proposed method. Results depict that the multiscale design cannot obviously improve structural stiffness compared with a solid-material design under the linear elastic condition.