基于DA算法的二维离散余弦逆变换的ASIC设计

　Aim　To present an ASIC design of DA-based 2-D IDCT. Methods In the design of 1-D IDCT is utilized a Chen-based fast IDCT algorithm, and multiplier accumulators based on distributed algorithm contributes in reducing the hardware amount and in enhancing the speed performance. Results and Conclusion VHDL simulation, synthesis and layout design of system are implemented. This 2-D IDCT ASIC design owns best timing performance when compared with other better designs internationally. Results of design prove to be excellent.     

动态系统的自适应模糊神经网络控制

　Aim　To build an adaptive fuzzy neural controller and simulate it. Methods Fuzzy logic and back propagation(BP) algorithm are combined to utilize their advantages while avoiding the disadvantages. Results and Conclusion Simulation results of the third-order plant with disturbances and dead times show the validity of the presented controller. The presented controller can control cases that preceding controllers were unable to control.     

不同衬底上低温生长GaN形貌的AFM观察

　Aim　To study the relationship between the substrate temperature and the morphology and properties of GaN. Methods Applying the hydride chemical vapor deposition method, GaN films were deposited on different kinds of substrates, including sapphire, Si(111),Si(100),GaAs and GaP(111) both on the P face and the Ga face. The growth was performed at low temperatures of below 700℃. XRD, Hall measurement, cathodoluminescence (CL) and atomic force microscopy (AFM) were used to characterise the film properties. Results It was found that the temperature and the nature of substrate materials influence the layer morphology. Conclusion The analysis shows that no apparent relationship exists between the optical properties and layer morphology.     

风险分析的尖点突变模型

　Aim　To assess simultaneously various risk states of a system. Methods Using the catastrophe and fuzzy theory, the energy and uncertainty in a system are set as two control variables and the function of the system is used as the state variable for analysis. Results and Conclusion A risk analysis model named the cusp model is presented. Various states regarding the safety of the system such as the accident state, no-accident state and miss state can be represented at will on the cusp model.     

用于车辆传动的金属材料的疲劳试验及统计方法

　Aim　To improve the efficiency of fatigue material tests and relevant statistical treatment of test data. Methods Least square approach and other special treatments were used. Results and Conclusion The concepts of each phase in fatigue tests and statistical treatment are clarified. The method proposed leads to three important properties. Reduced number of specimens brings to the advantage of lowering test expenditures. The whole test procedure has more flexibility for there is no need to conduct many tests at the same stress level as in traditional cases.     

极性反转瞬态电场的数值算法及应用

　Aim　To analyze the transient speciality of nonlinear, anisotropic, AC+DC coupling electric field, and to compare the withstand-voltage strength of different insulation structures. Methods The transient process of polarity reversal is analyzed, considering the anisotropic property of oil-immersed press-board, a new finite element model based on Galerkin method is presented and verified. The model developed is applied to calculate the electric field distribution in four typical winding end structures of the converter transformer. Results The whole-ring structure possesses the best insulation characteristics. Conclusion By introducing reasonable insulation components, insulation strength with the same surrounding sizes can be improved more than 30%.     

热处理对共混物中尼龙6熔融行为的影响

　Aim　To investigate the multiple melting behavior of polyamide-6(PA-6) in polyamide-6/linear low density polyethylene blends crystallized from the crystal-amorphous state. Methods The effects of annealing temperature, annealing time, heating rate, and the step-wise annealing were measured by DSC. Results and Conclusion There exists a critical heating rate affecting the middle temperature melting peak. When annealed at the temperature close to the melting peak, the main melting peak of PA-6 shifted to a higher temperature. Within a short time, annealing time has much effect on neat PA-6 but little effect on PA-6 in the blends. Addition of PE results in a decreasing in the height of melting peak. These phenomenon show that the melting behavior of PA-6 was affected by PE, compatibilizer, as well as thermal treatment.     

节理岩体弹塑性动态有限元分析

　Aim　To study the elastic-plastic dynamical constitutive relations about a jointed rock mass under explosion load and its computer simulation. Methods Stress history is taken into account and stresses will follow changes in time during a period of explosion load. According to the principle of static force balance, the corresponding nodal concentrated force is calculated and the nodal displacement is counted. The elastic-plastic dynamic finite element equations are thus obtained. Results A finite element method is given for a jointed rock mass under explosion load. Conclusion The problem of large plastic deformation for jointed rock mass on blasting was efficiently resolved through dynamic finite element analysis and the range of damages by blasting simulated, and this pushes forward the problem to engineering practice.     

燃料空气炸药近区抛散过程的研究

　Aim　To study fuel dispersion in fuel air explosive(FAE) and computational ways of fuel dispersion velocity in the near area. Methods The dispersion process of fuel in FAE was analyzed by the use of results measured with KODAK EKTAPRO EM Motion Analyzer and setting up mechanical models. Results Computational methods for fuel dispersion velocity in the acceleration stage is given and taken as a base for the study of fuel dispersion in the intermediate and the far area. Conclusion When the fuel flow velocity is higher than that of the explosion gas in the center cavity, the fuel divides with the explosion gas and its velocity of flow reaches a maximum. The acceleration stage ends at that time. The fuel dispersion velocity at this time is the initial conditions for numeral analyses of dispersion process in the intermediate and far areas.     

约束系统特性对乘员碰撞响应影响的仿真研究

　Aim　To study the influence of restraint system performance upon the occupant's response during impact, and provide a scientific base for occupant restraint system design. Methods In the light of basic theory of multibody system dynamics and impact dynamics on the basis of classical theory of impact, R-W method is adopted to construct the vehicle-occupant system model consisting of fourteen rigid bodies, thirty seven DOFs and slip joints for the simulation. A software named SVC3D(3-dimensional simulation of vehicle crash) is developed in the FORTRAN language. Results The results of simulation have a good coincidence with those of tests and the restraint system with low elongation webbing and equipped with pretensioner provides better restraint effect for the occupant. Conclusion The model of vehicle-occupant multibody system and SVC3D are suitable for use. Occupant should be belted with low elongation webbing to a certain degree and occupant restraint system should be equipped with pretensioner.     

正交多智能体算法求解管网直接优化调度模型

针对传统遗传算法存在高维空间寻优能力较差的问题,提出采用正交多智能体算法求解管网直接优化调度模型.对智能体的随机初始种群进行正交操作,得到较优初始种群,通过智能体的竞争和自学习,找到较优解.结果表明:与正交自适应遗传算法相比,正交多智能体算法具有更强全局搜索能力和更快寻优速度,并且正交多智能体算法优化方案较大提高了水泵运行效率,可节电2.96%.     

基于改进免疫遗传算法的网络优化设计

为了解决传统方法难以实现网络规划的组合优化问题, 采用改进免疫遗传算法对网络规划进行求解, 提高了种群的多样性和遗传算法的全局寻优能力。优化模型以网络架构费用最小为优化目标, 以同媒体施工、节点有限负载、网络辐射性连通为约束条件, 通过免疫规划进行网络结构的设计。进化实例表明, 改进免疫遗传算法与传统遗传算法相比, 具有较强的自适应能力和较好的效果。     

基于神经网络与遗传算法的结构优化设计方法

从神经网络和遗传算法的原理出发,利用遗传算法和神经网络相结合的策略对结构参数进行优化.在确定结构优化的目标函数和设计变量集合的基础上,用神经网络学习算法建立货架结构设计参数与结构重量、结构最大应力、最大位移等的非线性全局映射关系,获得遗传算法求解结构优化问题所需的目标函数,用遗传算法进行优胜劣汰的寻优搜索运算,从而求出所需最优解.以货架结构的优化为例说明了上述方法的应用.遗传算法和神经网络的优化结果是在正交设计法确定的训练样本足够大的基础上得出的,具有较强的可靠性.     

一种结合改进遗传算法和BP神经网络的图像压缩算法

针对LM（Levenberg—Marquardt）算法的缺陷,提出一种使用改进的遗传算法和LM算法优化神经网络的混合学习算法（GA-LMbp）。该算法先通过改进的遗传算法粗调得到一组全局最优近似解（即BP网络的初始权值和阈值）,再以该近似解为初值,用LM算法优化BP网络进行图像压缩处理。实验结果表明,新算法提高了网络的学习能力和收敛速度,避免了LMbp陷入平坦区或局部极小值。     

基于粒子群优化的频域多信道干扰对齐算法

针对频域干扰对齐系统解空间的多峰值特性,提出了一种基于粒子群优化,以系统网络和速率为优化目标函数的干扰对齐全局搜索算法.该算法通过对速度向量在位置向量的法平面上做投影以加强全局搜索能力,并在粒子群标准位置更新的基础上增加沿目标函数梯度方向的学习搜索来提高算法收敛速度和趋向全局最优值的能力.数值仿真结果表明,该算法可以获得比现有算法更好的网络和速率性能.     

基于双权值神经网络的煤气炉系统优化

一般的化工过程的输入输出系统具有高度的非线性特性,双权值神经网络具有较强的逼近与泛化能力,用工厂的实测数据,用双权值神经网络很好地逼近了一个煤气炉系统。利用遗传算法全局寻优的特点,找到了系统的若干组可行输入,并根据能耗评估,得到系统的最优输入。利用双权值神经网络及遗传算法为系统优化提供了一套解决方案。     

一种求解有约全局优化问题的新型混合方法

通过将模拟退火算法与非线性规划神经网络适当结合，本文提出一种求解有约束全局优化问题的新型混合方法．为了使该方法尽可能保持一般模拟退火算法通用性强的优点，在每一次迭代中不是采用非线性规划神经网络直接求原问题的局部最优解，而是通过求解一个辅助优化问题得到原问题的可行解．数值计算结果表明，与使用罚函数方法处理约束的模拟退火算法相比，本文提出的混合方法不仅可靠性高，而且可以显著地提高计算效率．     

动态变异遗传算法

遗传算法是根据达尔文生物进化理论而提出的一种优化算法。该文提出了一种新的遗传算法,理论分析显示,它不仅能保持遗传种群的多样性,而且能快速收敛。计算机仿真实验证明了改进后的遗传算法能够有效地克服不成熟收敛、进而搜索到全局最优解,并将这种新遗传算法用于BP网络的拓朴结构的优化和连接权值的训练,实例表明了该算法的有效性和可行性。     

基于遗传-神经网络算法的函数优化求解

提出了一种优化求解的遗传 神经网络算法 ,该算法综合遗传算法的全局性和神精网络算法的并行快速等特点 ,可克服传统数学方法不能精确求解、遗传算法最终进至最优解较慢和神经网络易陷入局部解的缺陷。实验结果表明具有全局性和收敛性     

基于小世界网络的PSO算法在电力系统中的应用

针对发电机的阀点效应,电力系统机组优化组合属于复杂的具有混合型变量的组合优化问题,基于小世界网络(Small World Network,SWN)的邻域模型构造,最优化原理和基本粒子群算法(ParticleSwarm Optimization,PSO),以24个时间段所有机组总耗量最小为目标函数,建立了发电机组优化组合的数学模型.提出了将"平均最短路径小,聚集系数大"的小世界网络邻域结构引入到粒子群算法中,以机组的输出功率作为粒子的位置,给出了算法的具体实现方法.在10机系统中分别采用了SWN-PSO算法和遗传算法进行了仿真计算.算例结果表明:所提出的算法不仅有利于粒子之间的信息共享,并且可以更快、更准确地收敛到全局最优解,具有一定的实用性.