直流输电线路故障行波波速不确定、波头提取困难以及噪声干扰等因素制约了直流电网中故障测距技术的应用。为了降低上述因素对定位准确性的影响,提出一种基于局部特征有理样条插值均值分解(LMD based on characteristic rational spline...直流输电线路故障行波波速不确定、波头提取困难以及噪声干扰等因素制约了直流电网中故障测距技术的应用。为了降低上述因素对定位准确性的影响,提出一种基于局部特征有理样条插值均值分解(LMD based on characteristic rational spline,CRS-LMD)和奇异值分解(singular value decomposition,SVD)的故障测距方法。首先,利用特征尺度选取最优极点系数,结合有理样条插值调节拟合曲线的松紧程度,实现对故障电压行波的局部均值分解。其次,采用奇异值分解对故障行波波头进行准确提取。最后,在PSCAD/EMTDC中搭建了张北±500 kV柔性直流电网的仿真模型,模拟各种故障情况并输出故障数据,利用Matlab对故障数据进行处理并验证定位算法。最后,仿真结果表明,所提故障测距算法在不同故障距离和故障类型下均能实现故障测距,且在叠加噪声和过渡电阻的情况下也能保障较高的精确性。展开更多
以无人驾驶小巴为代表的客运自动驾驶工具在公交微循环中发挥着重要作用,为实现无人驾驶技术的商业化规模运用,除了传统的车辆性能测试外,还需评估其在复杂场景下的表现。由于测试数据缺乏,评价模型适用场景单一和评价方法主观等问题,...以无人驾驶小巴为代表的客运自动驾驶工具在公交微循环中发挥着重要作用,为实现无人驾驶技术的商业化规模运用,除了传统的车辆性能测试外,还需评估其在复杂场景下的表现。由于测试数据缺乏,评价模型适用场景单一和评价方法主观等问题,导致以往评价偏差较大。本文针对无人驾驶小巴的表现构建综合评价体系,并在实测数据的基础上,采用博弈论组合赋权的优劣解距离法(Technique for Order Preference by Similarity to Ideal Solution,TOPSIS)模型,对车辆在复杂场景下的表现进行综合评价。选取驾驶安全性、乘坐舒适性、车辆智能性及车辆高效性这4个评价维度,并细分为12个客观评价指标。首先,通过实地测试采集无人驾驶小巴在运行场景中的数据;其次,利用基于博弈论的组合赋权法,对层次分析法和熵权法获得的权重进行组合;最后,为验证模型的有效性,运用TOPSIS模型对3条具有不同复杂度的测试路线进行综合评价值的计算。结果显示,无人驾驶小巴表现评价中,准则层的重要程度排序为车辆智能性、驾驶安全性、乘坐舒适性、车辆高效性,指标层敏感指标则为自动驾驶状态、平均角速度。基于博弈论组合赋权的TOPSIS模型对不同场景复杂度路线进行的无人驾驶小巴表现评价结果与实际运行情况一致,展示了方法的有效性。展开更多
With the maturation and advancement of blockchain technology,a novel execute-order-validate(EOV)architecture has been proposed,allowing transactions to be executed in parallel during the execution phase.However,parall...With the maturation and advancement of blockchain technology,a novel execute-order-validate(EOV)architecture has been proposed,allowing transactions to be executed in parallel during the execution phase.However,parallel execution may lead to multi-version concurrency control(MVCC)conflicts during the validation phase,resulting in transaction invalidation.Based on different causes,we categorize conflicts in the EOV blockchain into two types:within-block conflicts and cross-block conflicts,and propose an optimization solution called FabricMan based on Fabric v2.4.For within-block conflicts,a reordering algorithm is designed to improve the transaction success rate and parallel validation is implemented based on the transaction conflict graph.We also merge transfer transactions to prevent triggering multiple version checks.For cross-block conflicts,a cache-based version validation mechanism is implemented to detect and terminate invalid transactions in advance.Experimental comparisons are conducted between FabricMan and two other systems,Fabric and Fabric++.The results show that FabricMan outperforms the other two systems in terms of throughput,transaction abort rate,algorithm execution time,and other experimental metrics.展开更多
文摘针对无线传感网络中传统DV-Hop(Distance Vector Hop)定位算法节点分布不均匀导致定位误差较大的问题,提出了非均匀网络中半径可调的ARDV-Hop(Adjustable Radius DV-Hop in Non-uniform Networks)定位算法。该算法通过半径可调的方式对节点间的跳数进行细化,用细化后呈小数级的跳数代替传统的整数级跳数,并建立了数据能量消耗模型,优化了网络传输性能。ARDV-Hop算法还针对节点分布不均匀的区域提出跳距优化算法:在节点密度大的区域,采用余弦定理优化跳距;密度小的区域,采用最小均方误差(Least Mean Square,LMS)来修正跳距。仿真实验表明,在同等网络环境下,与传统DV-Hop算法、GDV-Hop算法和WOA-DV-Hop算法相比,ARDV-Hop算法能更有效地降低定位误差.
文摘直流输电线路故障行波波速不确定、波头提取困难以及噪声干扰等因素制约了直流电网中故障测距技术的应用。为了降低上述因素对定位准确性的影响,提出一种基于局部特征有理样条插值均值分解(LMD based on characteristic rational spline,CRS-LMD)和奇异值分解(singular value decomposition,SVD)的故障测距方法。首先,利用特征尺度选取最优极点系数,结合有理样条插值调节拟合曲线的松紧程度,实现对故障电压行波的局部均值分解。其次,采用奇异值分解对故障行波波头进行准确提取。最后,在PSCAD/EMTDC中搭建了张北±500 kV柔性直流电网的仿真模型,模拟各种故障情况并输出故障数据,利用Matlab对故障数据进行处理并验证定位算法。最后,仿真结果表明,所提故障测距算法在不同故障距离和故障类型下均能实现故障测距,且在叠加噪声和过渡电阻的情况下也能保障较高的精确性。
基金法国国家科研署(I'Agence nationale de la recherche)课题“联合国教科文组织与地方化的摩擦:全球治理下的遗产生产”(Unesco frictions:heritage-making across global governance,ANR-14-ACHN-0006-01)。
文摘以无人驾驶小巴为代表的客运自动驾驶工具在公交微循环中发挥着重要作用,为实现无人驾驶技术的商业化规模运用,除了传统的车辆性能测试外,还需评估其在复杂场景下的表现。由于测试数据缺乏,评价模型适用场景单一和评价方法主观等问题,导致以往评价偏差较大。本文针对无人驾驶小巴的表现构建综合评价体系,并在实测数据的基础上,采用博弈论组合赋权的优劣解距离法(Technique for Order Preference by Similarity to Ideal Solution,TOPSIS)模型,对车辆在复杂场景下的表现进行综合评价。选取驾驶安全性、乘坐舒适性、车辆智能性及车辆高效性这4个评价维度,并细分为12个客观评价指标。首先,通过实地测试采集无人驾驶小巴在运行场景中的数据;其次,利用基于博弈论的组合赋权法,对层次分析法和熵权法获得的权重进行组合;最后,为验证模型的有效性,运用TOPSIS模型对3条具有不同复杂度的测试路线进行综合评价值的计算。结果显示,无人驾驶小巴表现评价中,准则层的重要程度排序为车辆智能性、驾驶安全性、乘坐舒适性、车辆高效性,指标层敏感指标则为自动驾驶状态、平均角速度。基于博弈论组合赋权的TOPSIS模型对不同场景复杂度路线进行的无人驾驶小巴表现评价结果与实际运行情况一致,展示了方法的有效性。
文摘With the maturation and advancement of blockchain technology,a novel execute-order-validate(EOV)architecture has been proposed,allowing transactions to be executed in parallel during the execution phase.However,parallel execution may lead to multi-version concurrency control(MVCC)conflicts during the validation phase,resulting in transaction invalidation.Based on different causes,we categorize conflicts in the EOV blockchain into two types:within-block conflicts and cross-block conflicts,and propose an optimization solution called FabricMan based on Fabric v2.4.For within-block conflicts,a reordering algorithm is designed to improve the transaction success rate and parallel validation is implemented based on the transaction conflict graph.We also merge transfer transactions to prevent triggering multiple version checks.For cross-block conflicts,a cache-based version validation mechanism is implemented to detect and terminate invalid transactions in advance.Experimental comparisons are conducted between FabricMan and two other systems,Fabric and Fabric++.The results show that FabricMan outperforms the other two systems in terms of throughput,transaction abort rate,algorithm execution time,and other experimental metrics.
