模拟机器人足球比赛(Robot World Cup,RoboCup)作为多Agent系统的一个通用的实验平台,通过它可以来评价各种理论、算法和框架等,已经成为人工智能的研究热点。针对RoboCup仿真中的守门员防守问题,基于Q学习算法,描述了在特定场景中应用...模拟机器人足球比赛(Robot World Cup,RoboCup)作为多Agent系统的一个通用的实验平台,通过它可以来评价各种理论、算法和框架等,已经成为人工智能的研究热点。针对RoboCup仿真中的守门员防守问题,基于Q学习算法,描述了在特定场景中应用Q学习训练守门员的方法和过程。在RobCup中验证了该算法,实现了守门员防守策略的优化。展开更多
针对Robo Cup(Robot World Cup)中,多Agent之间的配合策略问题,采用了一种局部合作的多Agent Q-学习方法:通过细分球场区域和Agent回报值的方法,加强了Agent之间的协作能力,从而增强了队伍的进攻和防守能力。同时通过约束此算法的使用范...针对Robo Cup(Robot World Cup)中,多Agent之间的配合策略问题,采用了一种局部合作的多Agent Q-学习方法:通过细分球场区域和Agent回报值的方法,加强了Agent之间的协作能力,从而增强了队伍的进攻和防守能力。同时通过约束此算法的使用范围,减少了学习所用的时间,确保了比赛的实时性。最后在仿真2D平台上进行的实验证明,该方法比以前的效果更好,完全符合初期的设计目标。展开更多
As more and more variable frequency drives (VFDs), electronic ballasts, battery chargers, and static Var compensators are installed in facilities, the problems related to harmonics are expected to get worse. As a resu...As more and more variable frequency drives (VFDs), electronic ballasts, battery chargers, and static Var compensators are installed in facilities, the problems related to harmonics are expected to get worse. As a result Active power filter (APF) gains much more attention due to excellent harmonic compensation. But still the performance of the active filter seems to be in contradictions with different control strategies. This paper presents detailed analysis to compare and elevate the performance of two control strategies for ex-tracting reference currents of shunt active filters under balanced, un-balanced and non-sinusoidal conditions by using Fuzzy controller. The well known methods, instantaneous real active and reactive power method (p-q) and active and reactive current method (id-iq) are two control methods which are extensively used in active filters. Extensive Simulations are carried out with fuzzy controller for both p-q and Id-Iq methods for different voltage conditions and adequate results were presented. Simulation results validate the superior per-formance of active and reactive current control strategy (id-iq) with fuzzy controller over active and reactive power control strategy (p-q) with fuzzy controller.展开更多
插电式混合动力汽车(plug-in hybrid electric vehicles,PHEV)具有节能、环保、无续航里程焦虑的优点,是汽车领域发展的重点方向。但PHEV整车控制策略较为复杂,涉及到多动力源的能量分配,如何设计高效可靠的能量管理策略已经成为PHEV研...插电式混合动力汽车(plug-in hybrid electric vehicles,PHEV)具有节能、环保、无续航里程焦虑的优点,是汽车领域发展的重点方向。但PHEV整车控制策略较为复杂,涉及到多动力源的能量分配,如何设计高效可靠的能量管理策略已经成为PHEV研究的热点与难点。为了提升PHEV的燃油经济性和整车性能,提出了一种基于加权双Q学习的插电式混合动力汽车能量管理控制策略,采用加权双Q学习算法求解PHEV的能量分配。为了验证所提策略的有效性及可靠性,在Matlab/Simulink中搭建整车模型并进行仿真验证。研究结果表明:本文所提策略相比基于规则的CD/CS策略,燃油经济性在不同的行驶工况下平均提高6.38%;在不同的工况下,基于加权双Q学习策略的燃油经济性可达随机动态规划策略的98%,验证了本文所提策略具有较好的燃油经济性及工况适应性。展开更多
文摘模拟机器人足球比赛(Robot World Cup,RoboCup)作为多Agent系统的一个通用的实验平台,通过它可以来评价各种理论、算法和框架等,已经成为人工智能的研究热点。针对RoboCup仿真中的守门员防守问题,基于Q学习算法,描述了在特定场景中应用Q学习训练守门员的方法和过程。在RobCup中验证了该算法,实现了守门员防守策略的优化。
文摘针对Robo Cup(Robot World Cup)中,多Agent之间的配合策略问题,采用了一种局部合作的多Agent Q-学习方法:通过细分球场区域和Agent回报值的方法,加强了Agent之间的协作能力,从而增强了队伍的进攻和防守能力。同时通过约束此算法的使用范围,减少了学习所用的时间,确保了比赛的实时性。最后在仿真2D平台上进行的实验证明,该方法比以前的效果更好,完全符合初期的设计目标。
文摘As more and more variable frequency drives (VFDs), electronic ballasts, battery chargers, and static Var compensators are installed in facilities, the problems related to harmonics are expected to get worse. As a result Active power filter (APF) gains much more attention due to excellent harmonic compensation. But still the performance of the active filter seems to be in contradictions with different control strategies. This paper presents detailed analysis to compare and elevate the performance of two control strategies for ex-tracting reference currents of shunt active filters under balanced, un-balanced and non-sinusoidal conditions by using Fuzzy controller. The well known methods, instantaneous real active and reactive power method (p-q) and active and reactive current method (id-iq) are two control methods which are extensively used in active filters. Extensive Simulations are carried out with fuzzy controller for both p-q and Id-Iq methods for different voltage conditions and adequate results were presented. Simulation results validate the superior per-formance of active and reactive current control strategy (id-iq) with fuzzy controller over active and reactive power control strategy (p-q) with fuzzy controller.
文摘插电式混合动力汽车(plug-in hybrid electric vehicles,PHEV)具有节能、环保、无续航里程焦虑的优点,是汽车领域发展的重点方向。但PHEV整车控制策略较为复杂,涉及到多动力源的能量分配,如何设计高效可靠的能量管理策略已经成为PHEV研究的热点与难点。为了提升PHEV的燃油经济性和整车性能,提出了一种基于加权双Q学习的插电式混合动力汽车能量管理控制策略,采用加权双Q学习算法求解PHEV的能量分配。为了验证所提策略的有效性及可靠性,在Matlab/Simulink中搭建整车模型并进行仿真验证。研究结果表明:本文所提策略相比基于规则的CD/CS策略,燃油经济性在不同的行驶工况下平均提高6.38%;在不同的工况下,基于加权双Q学习策略的燃油经济性可达随机动态规划策略的98%,验证了本文所提策略具有较好的燃油经济性及工况适应性。