基于深度强化学习的智能船舶航迹跟踪控制

[目的]智能船舶的航迹跟踪控制问题往往面临着控制环境复杂、控制器稳定性不高以及大量的算法计算等问题。为实现对航迹跟踪的精准控制,提出一种引入深度强化学习技术的航向控制器。[方法]首先,结合视线(LOS)算法制导,以船舶的操纵特性和控制要求为基础,将航迹跟踪问题建模成马尔可夫决策过程,设计其状态空间、动作空间、奖励函数;然后,使用深度确定性策略梯度(DDPG)算法作为控制器的实现,采用离线学习方法对控制器进行训练;最后,将训练完成的控制器与BP-PID控制器进行对比研究,分析控制效果。[结果]仿真结果表明,设计的深度强化学习控制器可以从训练学习过程中快速收敛达到控制要求,训练后的网络与BP-PID控制器相比跟踪迅速,具有偏航误差小、舵角变化频率小等优点。[结论]研究成果可为智能船舶航迹跟踪控制提供参考。     

Development of knowledge based body structure concept design model

The purpose of this study is to propose a concept design process for an automotive body structure using technical information on the major joints and members of vehicles. First, in order to collect the technical information on major joints and members, 17 vehicles were selected using benchmark data. The collected technical information for the selected vehicles was the cross sectional shapes of each joint and member which were used for the analysis of joint stiffness, crashworthiness and static stiffness of the member to make a database along with cross section properties. This study applied a ‘What If Study’ technique to perform a concept design of an automotive body using the analyzed information and selected cross section meeting the design objectives. The criteria for the selection of the cross section were defined by subdividing the defined design objectives of an automotive body structure and constraints into members and joints. In order to configure an analysis model of an automotive body structure using the selected cross section, a shape parametric model was used and static stiffness, dynamic stiffness and crashworthiness were assessed to evaluate the configured automotive body structure. The evaluation result showed that the crashworthiness and static/dynamic stiffness were improved compared to an existing body structure. In addition, the weight of the body structure was reduced. Through this study, the process that can rapidly and effectively derive and evaluate the concept design of an automotive body structure was defined. It is expected that, henceforth, this process will be helpful for the study of automotive body structures.     

Nanoparticle emission characteristics and reduction strategies by boost pressure control and injection strategies in a passenger diesel engine

This research attempted to analyze nanoparticles and other harmful exhaust emissions in accordance with injection strategies and air-fuel ratio (AFR) changes for small diesel engines. The emission characteristics were analyzed in the medium-speed condition, which is the main driving range of a diesel engine. In the case of particulate matter (PM), the number of particles was measured, analyzed, and compared to identify the correlation and emission characteristics of nanoparticles by using a dilution device and condensation particle counter (CPC), which are international standards for particle measurement recommended by the Particulate Measurement Programme (PMP). The engine torque tended to be reduced as pilot injections were added, and the torque was increased by the increased boost pressure, but reduced by the exhaust pressure increase in a part of the low-load range. The number of nanoparticles was not influenced greatly by the change in AFR, but the reduction effect on the PM weight was great depending on the boost pressure increase. In addition, the number of nanoparticles tended to increase as the fuel injection timing became closer to TDC in all conditions, and its difference became larger with an increase in AFR. In addition, in the case of the pilot injection, nanoparticle emission showed similar characteristics depending on the main injection timing, but it was increased by advanced injection timing when performing the main injection only, and the number of the nanoparticles increased as pilot injections were added. Last, the optimal conditions for EMS calibration were analyzed by selecting the conditions of torque reduction and NOx increase within 5 % from all of the engine operating conditions; optimized conditions are presented.     

高速公路直线路段突起路标间隔设置方法

基于驾驶人的视觉惰性、视觉警示频率和感觉特性,分析了突起路标的视觉警示作用,确定了突起路标的闪现频率,计算与修正了突起路标的初始间隔。应用ADAMS/Car仿真软件模块,建立了道路、车辆和突起路标模型,将车型设置为大型车和小型车,大型车的车速设置为60、80、100km·h-1,小型车的车速设置为80、100、120km·h-1,大型车和小型车的方向盘转角均设置为1°、3°、5°。分别对提出的12m间隔与现行规范推荐的15m间隔各进行243次仿真,分析振动警示效果。仿真结果表明:在12m间隔下,平均碾压率为93.1%,小型车和大型车对应的平均警示率分别为41.7%和5.6%;在15m间隔下,平均碾压率为93.7%,小型车和大型车对应的平均警示率分别为33.3%和28.9%;12、15m间隔对小型车都具有较好的振动警示效果,15m间隔对大型车的振动警示效果更好。当高速公路夜间交通流较大或需要加大交通设施的夜间安全效果时,可选12m间隔提供良好的视觉连续性和视觉警示效果;考虑施工维护的方便性和经济性或者大型车比例较大时,可选择15m间隔。     

探究公路勘察设计与环境保护的关联

文章首先分析了公路建设过程中对环境造成的影响,而后阐述了公路工程在路线设计时应注意的环境问题,最后列举了提升公路设计线路与环境环保的举措,希望降低工程在建设过程中对环保造成的不利影响。     

瑞利面波椭圆极化在盾构隧道地层扰动探测中的应用

为解决传统物探方法在城市盾构隧道工程中难以对地层扰动达到较好探测和监测目的，提出了主动源瑞利面波椭圆极化方法. 采用二维有限元数值模拟对实际盾构隧道施工背景下的地层情况进行全波场正演模拟，并对其波场传播、速度频散和椭圆极化频散等特征进行分析；结合工程实例，进一步将椭圆极化方法和传统速度频散方法进行对比研究. 研究结果表明:主动源瑞利面波椭圆极化方法在探测地下介质结构分布及监测地下空洞等异常结构的变化方面具有和传统速度频散方法同样的探测效果，但由于其不需要长大检波器排列，对周围环境震动等噪声抗干扰能力相对较强，因此更适合城市等场地受限和具有一定震动和噪声干扰的环境，从而具有更广阔的应用前景.      

CA6123SH2型城市客车设计

简要介绍CA6123SH2客车的设计思想、整车造型、结构设计、工艺和主要技术参数。     

新型长寿命沥青路面结构力学指标的正交分析

针对我国高速公路多采用半刚性基层沥青路面结构,国外多为柔性基层结构,提出复合式基层结构。通过引进国外长寿命路面设计理念,提出新型长寿命沥青路面结构的力学设计指标;运用正交试验方法,设计了正交表L_(32)(4~9),通过极差分析得出影响力学设计指标的最不利因素,提出路面结构的最优组合。     

桥梁钢结构防腐涂装体系的选择

随着国家基础设施建设速度的加快,大型钢结构桥梁不断涌现。针对腐蚀和腐蚀疲劳问题而开展的钢结构防腐涂装研究日渐受到人们的关注。文中介绍了桥梁钢结构防腐的发展现状,论述了钢结构桥梁防腐涂装体系的选择和确定方法。