The Experts below are selected from a list of 13257 Experts worldwide ranked by ideXlab platform
Ivo Barbi - One of the best experts on this subject based on the ideXlab platform.
-
a Control strategy for parallel operation of single phase voltage source inverters analysis design and experimental results
IEEE Transactions on Industrial Electronics, 2013Co-Authors: Telles Brunelli Lazzarin, Guilherme A T Bauer, Ivo BarbiAbstract:This paper describes a theoretical and experimental study on a Control strategy for the parallel operation of single-phase voltage source inverters (VSI), to be applied to uninterruptible power supply. The Control system for each inverter consists of two main loops, which both use instantaneous values. The first (Parallelism Control) employs the feedback of the inductor current from the output filter to modify the input voltage of the same filter and, therefore, to Control the power flow of each inverter to the load. Additionally, the second loop (voltage Control) is responsible for Controlling the output voltage of the LC filter, which coincides with the output voltage of the VSI. Due to the fact that there is no exchange of information among the VSIs regarding their operation points, it is easier to obtain redundant systems. Furthermore, the connection (or disconnection) of inverters in a parallel arrangement is carried out directly, without connection impedance, and can occur at any operation point of the system. The proposed Control strategy ensures the proper sharing of the load current and avoids current circulation among the inverters during transient and steady-state operation. Moreover, its design and implementation are very simple. The Control technique was verified through experimental results with a maximum load of 10 kVA supplied by three parallel-connected inverters.
-
a Control strategy by instantaneous average values for parallel operation of single phase voltage source inverters based in the inductor current feedback
Energy Conversion Congress and Exposition, 2009Co-Authors: Telles Brunelli Lazzarin, Guilherme A T Bauer, Ivo BarbiAbstract:This paper presents a simple, effective and robust Control strategy applied to parallel connected of single phase voltage source inverters. The Parallelism Control is only done using internal measures of each inverter. The objective is to obtain a Control system where the inverters are independent from each other. The proposed Control strategy of each inverter has two Control loops by instantaneous average values: one for voltage and another for current. The first one is designed to Control the output voltage of the inverter, while the second is designed to lead the parallel operation of the inverters. The Parallelism Control acts in the voltage before the LC filter and it is based on the feedback of the L inductor current. Its function is to guarantee the correct load current sharing. The Control technique was corroborated through experimental results with a maximum load of 10 kVA supplied by three parallel connected inverters.
Telles Brunelli Lazzarin - One of the best experts on this subject based on the ideXlab platform.
-
a Control strategy for parallel operation of single phase voltage source inverters analysis design and experimental results
IEEE Transactions on Industrial Electronics, 2013Co-Authors: Telles Brunelli Lazzarin, Guilherme A T Bauer, Ivo BarbiAbstract:This paper describes a theoretical and experimental study on a Control strategy for the parallel operation of single-phase voltage source inverters (VSI), to be applied to uninterruptible power supply. The Control system for each inverter consists of two main loops, which both use instantaneous values. The first (Parallelism Control) employs the feedback of the inductor current from the output filter to modify the input voltage of the same filter and, therefore, to Control the power flow of each inverter to the load. Additionally, the second loop (voltage Control) is responsible for Controlling the output voltage of the LC filter, which coincides with the output voltage of the VSI. Due to the fact that there is no exchange of information among the VSIs regarding their operation points, it is easier to obtain redundant systems. Furthermore, the connection (or disconnection) of inverters in a parallel arrangement is carried out directly, without connection impedance, and can occur at any operation point of the system. The proposed Control strategy ensures the proper sharing of the load current and avoids current circulation among the inverters during transient and steady-state operation. Moreover, its design and implementation are very simple. The Control technique was verified through experimental results with a maximum load of 10 kVA supplied by three parallel-connected inverters.
-
a Control strategy by instantaneous average values for parallel operation of single phase voltage source inverters based in the inductor current feedback
Energy Conversion Congress and Exposition, 2009Co-Authors: Telles Brunelli Lazzarin, Guilherme A T Bauer, Ivo BarbiAbstract:This paper presents a simple, effective and robust Control strategy applied to parallel connected of single phase voltage source inverters. The Parallelism Control is only done using internal measures of each inverter. The objective is to obtain a Control system where the inverters are independent from each other. The proposed Control strategy of each inverter has two Control loops by instantaneous average values: one for voltage and another for current. The first one is designed to Control the output voltage of the inverter, while the second is designed to lead the parallel operation of the inverters. The Parallelism Control acts in the voltage before the LC filter and it is based on the feedback of the L inductor current. Its function is to guarantee the correct load current sharing. The Control technique was corroborated through experimental results with a maximum load of 10 kVA supplied by three parallel connected inverters.
Guilherme A T Bauer - One of the best experts on this subject based on the ideXlab platform.
-
a Control strategy for parallel operation of single phase voltage source inverters analysis design and experimental results
IEEE Transactions on Industrial Electronics, 2013Co-Authors: Telles Brunelli Lazzarin, Guilherme A T Bauer, Ivo BarbiAbstract:This paper describes a theoretical and experimental study on a Control strategy for the parallel operation of single-phase voltage source inverters (VSI), to be applied to uninterruptible power supply. The Control system for each inverter consists of two main loops, which both use instantaneous values. The first (Parallelism Control) employs the feedback of the inductor current from the output filter to modify the input voltage of the same filter and, therefore, to Control the power flow of each inverter to the load. Additionally, the second loop (voltage Control) is responsible for Controlling the output voltage of the LC filter, which coincides with the output voltage of the VSI. Due to the fact that there is no exchange of information among the VSIs regarding their operation points, it is easier to obtain redundant systems. Furthermore, the connection (or disconnection) of inverters in a parallel arrangement is carried out directly, without connection impedance, and can occur at any operation point of the system. The proposed Control strategy ensures the proper sharing of the load current and avoids current circulation among the inverters during transient and steady-state operation. Moreover, its design and implementation are very simple. The Control technique was verified through experimental results with a maximum load of 10 kVA supplied by three parallel-connected inverters.
-
a Control strategy by instantaneous average values for parallel operation of single phase voltage source inverters based in the inductor current feedback
Energy Conversion Congress and Exposition, 2009Co-Authors: Telles Brunelli Lazzarin, Guilherme A T Bauer, Ivo BarbiAbstract:This paper presents a simple, effective and robust Control strategy applied to parallel connected of single phase voltage source inverters. The Parallelism Control is only done using internal measures of each inverter. The objective is to obtain a Control system where the inverters are independent from each other. The proposed Control strategy of each inverter has two Control loops by instantaneous average values: one for voltage and another for current. The first one is designed to Control the output voltage of the inverter, while the second is designed to lead the parallel operation of the inverters. The Parallelism Control acts in the voltage before the LC filter and it is based on the feedback of the L inductor current. Its function is to guarantee the correct load current sharing. The Control technique was corroborated through experimental results with a maximum load of 10 kVA supplied by three parallel connected inverters.
Yanfei Sun - One of the best experts on this subject based on the ideXlab platform.
-
falcon addressing stragglers in heterogeneous parameter server via multiple Parallelism
IEEE Transactions on Computers, 2021Co-Authors: Qihua Zhou, Song Guo, Minyi Guo, Yanfei Sun, Kun WangAbstract:The parameter server architecture has shown promising performance advantages when handling deep learning (DL) applications. One crucial issue in this regard is the presence of stragglers, which significantly retards DL training progress. Previous solutions for solving stragglers may not fully exploit the computation resource of the cluster as evidenced by our experiments, especially in the heterogeneous environment. This motivates us to design a heterogeneity-aware parameter server paradigm that addresses stragglers and accelerates DL training from the perspective of computation Parallelism. We introduce a novel methodology named straggler projection to give a comprehensive inspection of stragglers and reveal practical guidelines to solve this problem in two aspects: (1) Controlling each worker's training speed via elastic training Parallelism Control and (2) transferring blocked tasks from stragglers to pioneers to fully utilize the computation resource. Following these guidelines, we propose the abstraction of Parallelism as an infrastructure and design the Elastic-Parallelism Synchronous Parallel (EPSP) algorithm to handle distributed training and parameter synchronization, supporting both enforced- and slack-synchronization schemes. The whole idea has been implemented into a prototype called ${\sf Falcon}$ Falcon which effectively accelerates the DL training speed with the presence of stragglers. Evaluation under various benchmarks with baseline comparison demonstrates the superiority of our system. Specifically, ${\sf Falcon}$ Falcon reduces the training convergence time, by up to 61.83, 55.19, 38.92, and 23.68 percent shorter than FlexRR, Sync-opt, ConSGD, and DynSGD, respectively.
-
falcon towards computation parallel deep learning in heterogeneous parameter server
International Conference on Distributed Computing Systems, 2019Co-Authors: Qihua Zhou, Kun Wang, Song Guo, Minyi Guo, Yanfei SunAbstract:Parameter server paradigm has shown great performance superiority for handling deep learning (DL) applications. One crucial issue in this regard is the presence of stragglers, which significantly retards DL training progress. Previous solutions for solving straggler may not fully exploit the computation capacity of a cluster as evidenced by our experiments. This motivates us to make an attempt at building a new parameter server architecture that mitigates and addresses stragglers in heterogeneous DL from the perspective of computation Parallelism. We introduce a novel methodology named straggler projection to give a comprehensive inspection of stragglers and reveal practical guidelines for resolving this problem: (1) reducing straggler emergence frequency via elastic Parallelism Control and (2) transferring blocked tasks to pioneer workers for fully exploiting cluster computation capacity. Following the guidelines, we propose the abstraction of Parallelism as an infrastructure and elaborate the Elastic-Parallelism Synchronous Parallel (EPSP) that supports both enforced-and slack-synchronization schemes. The whole idea has been implemented in a prototype called Falcon which efficiently accelerates the DL training progress with the presence of stragglers. Evaluation under various benchmarks with baseline comparison evidences the superiority of our system. Specifically, Falcon yields shorter convergence time, by up to 61.83%, 55.19%, 38.92% and 23.68% reduction over FlexRR, Sync-opt, ConSGD and DynSGD, respectively.
Qihua Zhou - One of the best experts on this subject based on the ideXlab platform.
-
falcon addressing stragglers in heterogeneous parameter server via multiple Parallelism
IEEE Transactions on Computers, 2021Co-Authors: Qihua Zhou, Song Guo, Minyi Guo, Yanfei Sun, Kun WangAbstract:The parameter server architecture has shown promising performance advantages when handling deep learning (DL) applications. One crucial issue in this regard is the presence of stragglers, which significantly retards DL training progress. Previous solutions for solving stragglers may not fully exploit the computation resource of the cluster as evidenced by our experiments, especially in the heterogeneous environment. This motivates us to design a heterogeneity-aware parameter server paradigm that addresses stragglers and accelerates DL training from the perspective of computation Parallelism. We introduce a novel methodology named straggler projection to give a comprehensive inspection of stragglers and reveal practical guidelines to solve this problem in two aspects: (1) Controlling each worker's training speed via elastic training Parallelism Control and (2) transferring blocked tasks from stragglers to pioneers to fully utilize the computation resource. Following these guidelines, we propose the abstraction of Parallelism as an infrastructure and design the Elastic-Parallelism Synchronous Parallel (EPSP) algorithm to handle distributed training and parameter synchronization, supporting both enforced- and slack-synchronization schemes. The whole idea has been implemented into a prototype called ${\sf Falcon}$ Falcon which effectively accelerates the DL training speed with the presence of stragglers. Evaluation under various benchmarks with baseline comparison demonstrates the superiority of our system. Specifically, ${\sf Falcon}$ Falcon reduces the training convergence time, by up to 61.83, 55.19, 38.92, and 23.68 percent shorter than FlexRR, Sync-opt, ConSGD, and DynSGD, respectively.
-
falcon towards computation parallel deep learning in heterogeneous parameter server
International Conference on Distributed Computing Systems, 2019Co-Authors: Qihua Zhou, Kun Wang, Song Guo, Minyi Guo, Yanfei SunAbstract:Parameter server paradigm has shown great performance superiority for handling deep learning (DL) applications. One crucial issue in this regard is the presence of stragglers, which significantly retards DL training progress. Previous solutions for solving straggler may not fully exploit the computation capacity of a cluster as evidenced by our experiments. This motivates us to make an attempt at building a new parameter server architecture that mitigates and addresses stragglers in heterogeneous DL from the perspective of computation Parallelism. We introduce a novel methodology named straggler projection to give a comprehensive inspection of stragglers and reveal practical guidelines for resolving this problem: (1) reducing straggler emergence frequency via elastic Parallelism Control and (2) transferring blocked tasks to pioneer workers for fully exploiting cluster computation capacity. Following the guidelines, we propose the abstraction of Parallelism as an infrastructure and elaborate the Elastic-Parallelism Synchronous Parallel (EPSP) that supports both enforced-and slack-synchronization schemes. The whole idea has been implemented in a prototype called Falcon which efficiently accelerates the DL training progress with the presence of stragglers. Evaluation under various benchmarks with baseline comparison evidences the superiority of our system. Specifically, Falcon yields shorter convergence time, by up to 61.83%, 55.19%, 38.92% and 23.68% reduction over FlexRR, Sync-opt, ConSGD and DynSGD, respectively.
