Browsing by Author "Amarasinghe, G"

Now showing 1 - 4 of 4

item: Conference-Full-text
Application of novel evolutionary algorithms for analyzing the impact of integrating renewables on the adequacy of composite power systems
(IEEE, 2022-07) Amarasinghe, G; Abeygunawardane, S; Singh, C; Rathnayake, M; Adhikariwatte, V; Hemachandra, K
The adequacy evaluation of modern renewable-rich power systems tends to be a computationally challenging task due to variations of renewable power generation. Recently, more computationally efficient evolutionary algorithms and swarm intelligence-based methods are utilized for evaluating the adequacy of power systems. In this paper, the authors have proposed a wind and solar integrated composite system adequacy evaluation framework using an Evolutionary Swarm Algorithm (ESA). The system failure states which have a higher probability of occurrence are explored using the ESA to estimate the adequacy indices of the system. The wind and solar power generation are modeled using a clustering-based method considering their annual effective power output. Moreover, the correlation between the system load and renewable power generation is modeled in the adequacy evaluation framework. Using the proposed framework, several case studies are conducted on the IEEE Reliability Test System to analyze the impact of integrating renewables on the adequacy of composite systems. The results show that wind generation tends to improve system reliability than solar due to its higher availability. In addition, the equivalent capacities of wind and solar generators are found to be 125MW and 215MW against a 50MW hydro generator.
item: Conference-Abstract
Detection of false sharing using machine learning
(2014-06-25) Jayasena, VSD; Amarasinghe, S; Abeyweera, A; Amarasinghe, G; De Silva, H; Rathnayake, S; Meng, X; Liu, Y
False sharing is a major class of performance bugs in parallel applications. Detecting false sharing is difficult as it does not change the program semantics. We introduce an efficient and effective approach for detecting false sharing based on machine learning. We develop a set of mini-programs in which false sharing can be turned on and off. We then run the mini-programs both with and without false sharing, collect a set of hardware performance event counts and use the collected data to train a classifier. We can use the trained classifier to analyze data from arbitrary programs for detection of false sharing. Experiments with the PARSEC and Phoenix benchmarks show that our approach is indeed effective. We detect published false sharing regions in the benchmarks with zero false positives. Our performance penalty is less than 2%. Thus, we believe that this is an effective and practical method for detecting false sharing.
item:
Detection of false sharing using machine learning
(2015-06-19) Jayasena, S; Amarasinghe, S; Abeyweera, A; Amarasinghe, G; De Silva, H; Rathnayake, S; Meng, X; Liu, Y
False sharing is a major class of performance bugs in parallel applications. Detecting false sharing is difficult as it does not change the program semantics. We introduce an efficient and effective approach for detecting false sharing based on machine learning. We develop a set of mini-programs in which false sharing can be turned on and off. We then run the mini-programs both with and without false sharing, collect a set of hardware performance event counts and use the collected data to train a classifier. We can use the trained classifier to analyze data from arbitrary programs for detection of false haring. Experiments with the PARSEC and Phoenix benchmarks show that our approach is indeed effective. We detect published false sharing regions in the benchmarks with zero false positives. Our performance penalty is less than 2%. Thus, we believe that this is an effective and practical method for detecting false sharing.
item: Conference-Full-text
Short-term wind power forecasting using a Markov model
(Institute of Electrical and Electronics Engineers, Inc., 2021-09) Jeyakumar, P; Kolambage, N; Geeganage, N; Amarasinghe, G; Abeygunawardane, SK; Abeykoon, AMHS; Velmanickam, L
Large-scale wind power integration to power systems has been significantly increasing since the last decade. However, the reliability of power systems tends to degrade due to the intermittency and uncontrollability of wind power. Future wind power generation forecasts can be used to reduce the impacts of intermittency and uncontrollability of wind power on the reliability of power systems. This paper proposes a Markov chain-based model for the short-term forecasting of wind power. The first-order and second-order Markov chain principles are used as they require lesser memory and have lower complexities. Seasonal variation is also incorporated into the proposed model to further improve the accuracy. Results obtained from both Markov models are validated with real wind power output data and evaluated using evaluation metrics such as Mean Square Error and Root Mean Square Error. The results show that the accuracy of the first-order and second-order Markov models for a high wind regime is 81.33% and 82.61%, respectively and for a low wind regime is 83.50% and 89.27% respectively.