压力容器爆破试验实验装置的设计与探讨

“压力容器爆破试验”实验的教学要求是：(1)能测定“压力——膨胀量”关系曲线;(2)能测定试验容器爆破压力;(3)爆破后能观察爆破口的形貌;(4)实验后便于恢复原始状态,适合教学反复使用等。根据上述要求,本文对“压力容器爆破试验”的试验容器、实验装置进行了设计与探讨。     

应用化学专业实践教学模式初探

介绍了以江西农业大学理学院基础化学实验室为基础,以2002级应化专业学生为对象,以培养具有“基础扎实、知识面广、能力强、素质高”的优秀化学人才为目标,改革应用化学专业实践教学内容和方法,改善实验条件.     

180°矩形弯管流场的实验测量和数值模拟

对180°矩形弯管内的湍流流场用智能型五孔球探针进行了测量，同时采用FLUENT 6.2对其进行数值模拟。流场测量和数值模拟结果表明：180°矩形弯管由于流体旋转产生的离心力导致了弯管内压力分布的变化，使弯管内切向速度发生变化。切向速度在流体开始旋转0°~60°区间内侧的速度增大、压力减小，外侧的速度降低、压力增大；当转过60°截面后，外侧的速度增大、压力减小，内侧的速度降低、压力增大。     

A unified model for the structure of polymers in semidilute solution

A model is developed to analyse the concentration dependence of the range ξ_? of the monomer pair correlation function. In semidilute solution, three concentration regimes are found for semiflexible molecules and the crossover points between the various regimes are predicted in terms of the characteristic ratio of the chains in dilute solution and in terms of the Flory interaction parameter χ. A simple physical interpretation is given which explains the concentration dependence of ξ_? based on binary contacts initially and then ternary contacts at higher concentration. Temperature-concentration diagrams are developed for several common polymer-solvent systems.     

C4烃催化裂解动力学装置进料方法的研究

研究了液态进料法、气态进料法等不同进料方法对C4烯烃催化裂解动力学研究装置物料平衡的影响。结果表明，气态间接进料法可满足动力学研究的需要，并建立了稳定可靠的动力学研究装置。     

集散控制装置的实验功能开发

以一个模拟生产设备为监控对象,介绍了PCS型集散控制装置的硬件和软件组成,并开发了该装置的各种实验功能,可满足实验教学的要求。     

聚合物气辅共注成型工艺的实验研究

针对气辅共注成型中多相分层流动存在着各分层界面应力间的相互耦合,使得其成型过程具有特殊的流动输运规律和动力学特征,对气辅共注成型工艺进行了系统的实验研究。通过实验研究了熔体注射温度、气体填充量和气体延迟时间3种工艺参数对气辅共注成型的影响规律,并在此基础上通过理论分析揭示了工艺参数对气辅共注成型的影响机理。     

THROO: a Tracked Hybrid Rover to Overpass Obstacles

In this paper, the design, simulation and experimental tests are presented for THROO: a Tracked Hybrid Rover, which has been developed to Overpass Obstacles. The proposed mobile robot has 3-DOFs and it is capable of straight motion, turning ability and two operations, namely rover-like motion with tracks on flat terrain and walking-like motion with track and legs to overpass obstacles to move on uneven terrain. The leg mechanism is composed of a four-bar linkage, which has been synthesized according to the desired features. In particular, the Burmester problem, which aims at finding the geometric parameters of a four-bar linkage required for a prescribed set of finitely separated poses are solved for the case understudy. Dynamic simulations have been carried out and a prototype has been built. The proposed results show the hybrid rover ability to overpass obstacles, for which size is comparable or greater than the track high.     

基于中心复合设计的振荡热管传热性能分析

Oscillating heat pipes （OHPs） are very promising cooling devices. Their heat transfer performance is affected by many factors, and the form of the relationship between the performance and the factors is complex and non-linear. In this paper, the effects of charging ratio, inclination angle, and heat input and their interaction effects on heat transfer performance of a looped copper-water OHP are analyzed. First, suppose that the relationship between the response and the variables approximates a second-order model. And use the central composite design to arrange the experiment. Then, the method of least squares is used to estimate the parameters in the second-order model. Finally, multivariate variance analysis is used to analyze the model. The results show that the assumption is right, that is to say, the re- lationship is well modeled by a second-order function. Among the three main effect variables, the effect of inclination angle is the most significant, but their interaction effects are not significant. In the range of the considered factors, both the optimum charging ratio and the optimum inclination angle increase as the heating water flow rate increases.     

多标记分类和标记相关性的联合学习

提出了多标记分类和标记相关性的联合学习(JMLLC),在JMLLC中,构建了基于类别标记变量的有向条件依赖网络,这样不仅使得标记分类器之间可以联合学习,从而增强各个标记分类器的学习效果,而且标记分类器和标记相关性可以联合学习,从而使得学习得到的标记相关性更为准确.通过采用两种不同的损失函数:logistic回归和最小二乘,分别提出了JMLLC-LR(JMLLC with logistic regression)和JMLLC-LS(JMLLC with least squares),并都拓展到再生核希尔伯特空间中.最后采用交替求解的方法求解JMLLC-LR和JMLLC-LS.在20个基准数据集上基于5种不同的评价准则的实验结果表明,JMLLC优于已提出的多标记学习算法.