基于粒子群算法的隐式广义预测在ATO中的应用

由于列车运行速度的不断提高,对列车自动驾驶(Automatic Train Operation,ATO)系统提出更高的要求。针对隐式广义预测(Implicit Generalized Predictive Control,IGPC)控制器在ATO中难以获得最优预测控制输入的问题,运用一种基于粒子群优化(Particle Swarm Optimization,PSO)的IGPC算法对ATO系统进行控制。为更进一步提高PSO算法的寻优能力,对基本PSO算法进行改进,从而有效提高系统的寻优精度和速度。并对有约束情况下的CRH2型车进行仿真验证,仿真结果显示PSO-IGPC比单纯IGPC对ATO的控制效果更优。     

基于铁路无线通信环境下改进的PSO算法在多用户检测中的应用

随着无线通信技术在铁路系统中迅猛发展和广泛应用,对通信系统装备的抗干扰能力提出更高更新的要求,因此必须研究有效的抗干扰技术以对付日益严重的干扰威胁。在粒子群算法中采用量子力学和经典力学的理论,将量子离散粒子群算法应用于多用户检测技术。在铁路无线通信系统中,可以提高铁路信道质量,降低数据通信误码率,解决列车与调度指挥中心之间的高数据通信中误码率高的问题,并且在抗多址干扰能力明显优于传统检测器。     

粒径对块石层自然对流特性影响的试验研究

用粒径为5～8,8～15和15～20 cm的3种块石分别填筑成块石层模型,在实验室内研究它们的空气自然对流特性。结果表明,在块石层模型下表面保持20℃而上表面从20℃降至-20℃的过程中,模型内均产生了Benard流,而且模型内空气自然对流的强弱与块石的粒径有关,粒径为15～20 cm的块石层模型内气流速度最大,约为2.2 m.s-1;粒径为8～15 cm的最小,约为1.5 m.s-1,这种变化主要是由块石层内热弥散效应以及块石与空气间接触面积的改变而引起的。控温方式对空气自然对流影响的研究表明,由模型上表面降温所产生的空气自然对流效应强于在块石层模型下表面升温,故在实际工程中,冬季块石层内的空气自然对流更易于发生。针对模型内空气流速计算值均比实测值小的问题,建议在进行多孔块石结构对流换热理论计算时,应考虑块石层模型的假设条件、热弥散作用、流动惯性以及空气热物性随压力的变化等的影响。     

PSO algorithm-based parameter optimization for HEV powertrain and its control strategy

The coordination between the powertrain and control strategy has significant impacts on the operating performance of hybrid electric vehicles (HEVs). A comprehensive methodology based on Particle Swarm Optimization (PSO) is presented in this paper to achieve parameter optimization for both the powertrain and the control strategy, with the aim of reducing fuel consumption, exhaust emissions, and manufacturing costs of the HEV. The original multi-objective optimization problem is converted into a single-objective problem with a goal-attainment method, and the principal parameters of powertrain and control strategy are set as the optimized variables by PSO, with the dynamic performance index of HEVs being defined as the constraint condition. Computer simulations were carried out, which showed that the PSO scheme gives preferable results in comparison to the ADVISOR method. Therefore, fuel consumption and exhaust emissions of HEVs can be effectively reduced without sacrificing dynamic performance of HEVs.     

Three-dimensional numerical analysis of shipping water onto a moving ship using a particle method

The objective of this study was to develop a numerical analysis method based on the moving particle semi-implicit method for simulating shipping water on a moving ship. Towing tests of a very large crude carrier were numerically analyzed for three typical wavelengths. The ship was forced to move in order to express previously measured ship oscillations, and the calculated fluid behavior and the impact pressure on the deck were compared with the experimental results.     

基于粒子群优化小波神经网络的行程时间预测

为使道路使用者在出发前获得具有高实时性和可靠性的行程时间预测信息,提高出行效率,需提升高速公路行程时间的预测精度。鉴于此,将生物学中粒子群优化算法（Particle Swarm Optimization Algorithm, PSO）引入小波神经网络（Wavelet Neural Network, WNN）中,构建基于粒子群优化小波神经网络（Particle Swarm Optimization Wavelet Neural Network, PSO-WNN）的高速公路行程时间预测模型。首先将高速公路原始收费数据规整化,截取其中有效字段,获取研究路段一个月的行程时间数据并对其进行数据处理。然后分别基于PSO-WNN模型和WNN模型,利用Matlab进行实验。实验结果显示,PSO-WNN模型预测结果的平均绝对误差、平均相对误差和均方误差较WNN模型分别降低了83.36%, 82.20%和98.15%。PSO-WNN行程时间预测模型不仅预测精度高,而且能较准确地预测出行程时间的走向及波动情况,在收敛速度方面也呈现出一定的优势,具有较好的适应能力。     

Personal exposures to PM during short distance highway travel in India

Traffic is the largest contributor (37%) to urban air pollution in India. During commuting, passengers are significantly exposed to pollutants. We carried out a study on a National Highway (NH) in India to measure personal exposure to Particulate Matter (PM) in five travel modes. PM_2.5 concentrations showed the following trend: Bus > Car FA (fresh air mode of air condition) > Bus AC > Car > Car RC (re-circulation mode of air condition). Highest and lowest concentrations of PM₁₀ were observed in Bus (134 ± 47 µg m⁻³) and Car RC (20 ± 5 µg m⁻³), respectively. The exposures were highest at the rear seats during the Bus AC journeys. In Car FA, the contribution of PM₁ to total concentrations was dominant (61%). Travel modes explained highest variabilities in PM₁₀, PM_2.5 and PM₁ concentrations. In all travel modes, the highest particle counts were observed for PM_0.3–0.5. PM_>0.5–5.0 counts during Bus journeys were comparatively higher than remaining modes. Deposition doses of passengers were as high as 3.22 µg of PM₁₀ (in Bus), 0.66 µg of PM_2.5 (in Bus) and 0.06 µg of PM₁ (in Bus AC) during the ~1 h journey. Our study revealed that Car RC is the safest mode of travel, both in terms of personal exposures and PM depositions in respiratory system. The results from this study can be used to target efforts to reduce personal exposure of highway commuters.     

柴油-甲缩醛混合燃料喷雾的粒子动态分析研究

为研究代用燃料甲缩醛的喷雾机理,采用基于相位多普勒原理的粒子动态分析仪(PDA),研究了不同百分比的柴油-甲缩醛混合燃料喷雾的粒度场、速度场。试验结果表明,柴油添加甲缩醛后,雾化得到改善,喷雾轴线上的粒子轴向速度提高,但三者径向速度相差不大。     

颗粒阻尼应用在建筑上试验初探

分析了颗粒阻尼技术应用于建筑结构可能性,并指出了颗粒阻尼在建筑物上应用的具体位置和安装方式;通过单层和多层钢框架模型的振动台试验,探讨了砂子、钢珠等材料作为建筑结构颗粒阻尼材料的减振性能和减振效果。实验表明:粗砂作为建筑结构的颗粒阻尼材料,可以提供较大的等效阻尼比,能够明显减轻结构的大震反应。     

基于IPSO-Capsule-NN模型的中欧班列出口需求量预测

为提高中欧班列出口需求量的预测精度,提出将改进粒子群算法(IPSO)与胶囊神经网络(Capsule-NN)相结合的预测模型(IPSO-Capsule-NN)。与全连接神经网络不同,胶囊神经网络通过动态路由算法增强了模型的拟合能力和泛化能力。利用改进粒子群算法优化胶囊神经网络的神经元数量、迭代次数以及学习率,以克服人为设定模型参数随机性较大导致模型精确度不高的不足之处。此外,针对标准粒子群算法存在的缺点,提出一种非线性递减惯性权重并引入Levy飞行对粒子群算法的全局寻优能力和收敛速度进行优化。将采用spearman秩相关性分析得到的11个因素作为中欧班列出口需求量的影响因素并对其进行预测,结果表明:胶囊神经网络具有2层隐含层时,IPSO-Capsule-NN模型预测精度更高。