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
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
为保证有限控制集模型预测控制(finite control set model predictive control,FCS-MPC)在功率变换器运行效果,其采样频率一般设定为10~50 kHz。但由于会出现连续多个采样周期输出同一开关状态现象,实际开关频率仅为采样频率的10%~20%,...为保证有限控制集模型预测控制(finite control set model predictive control,FCS-MPC)在功率变换器运行效果,其采样频率一般设定为10~50 kHz。但由于会出现连续多个采样周期输出同一开关状态现象,实际开关频率仅为采样频率的10%~20%,影响控制精度。在SiC、GaN等高速器件应用场景中,一味地提升采样频率实现高开关频率输出,将对数字处理器造成极大地运算负担。针对上述问题,提出一种多速率模型预测控制方法(multi-rate model predictive control,MR-MPC)方法,通过构建低采样输入、高控制输出的双速率MR-MPC离散预测模型,并引入多层次递归优化技术,实现计算负担和控制性能之间平衡。最后,搭建5 kW电机测试平台进行实验验证与分析,MR-MPC继承了传统FCS-MPC处理复杂控制目标的能力,并可在低运算负担条件下,实现输出开关频率多速率倍频,且倍频比例越高,系统动稳态性能越优,不失为高开关频率功率变换器通用型设计方法之一。展开更多
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.
文摘为保证有限控制集模型预测控制(finite control set model predictive control,FCS-MPC)在功率变换器运行效果,其采样频率一般设定为10~50 kHz。但由于会出现连续多个采样周期输出同一开关状态现象,实际开关频率仅为采样频率的10%~20%,影响控制精度。在SiC、GaN等高速器件应用场景中,一味地提升采样频率实现高开关频率输出,将对数字处理器造成极大地运算负担。针对上述问题,提出一种多速率模型预测控制方法(multi-rate model predictive control,MR-MPC)方法,通过构建低采样输入、高控制输出的双速率MR-MPC离散预测模型,并引入多层次递归优化技术,实现计算负担和控制性能之间平衡。最后,搭建5 kW电机测试平台进行实验验证与分析,MR-MPC继承了传统FCS-MPC处理复杂控制目标的能力,并可在低运算负担条件下,实现输出开关频率多速率倍频,且倍频比例越高,系统动稳态性能越优,不失为高开关频率功率变换器通用型设计方法之一。