Pub Date : 2018-04-12DOI: 10.1109/AGENTS.2018.8460096
Y. Yoshida, S. Arai
From the viewpoint of energy conservation in the railway systems, the effective usage of regenerative power generated during train braking has attracted a lot of attention lately. To utilize regenerative power with balancing the electric power supply-demand, we introduce a storage battery, and propose a charge control method of it. Our proposed algorithm could make not only balance the electric power supply-demand but also suppresses the fluctuation of the charged amount within the storage battery. The smaller amount of charge fluctuation, the smaller capacity battery would be available to use. In several existing methods, the empirical rules have been adopted to secure the balance, without consideration for suppressing the fluctuations of charged amount electricity. However, rule-based control which is based on the human empirical knowledge, has some limitations in electricity supply-demand dynamics in the railway systems. To overcome the limitations, we introduce reinforcement learning with an actor-critic algorithm to acquire the effective control policy which had been difficult to draw from the experts' knowledge as the rules. Through several computational simulations, we verified that the performance of our proposed method shows superior to that of the existing one.
{"title":"Charge Control of Regenerative Power for Energy Saving in Railway Systems","authors":"Y. Yoshida, S. Arai","doi":"10.1109/AGENTS.2018.8460096","DOIUrl":"https://doi.org/10.1109/AGENTS.2018.8460096","url":null,"abstract":"From the viewpoint of energy conservation in the railway systems, the effective usage of regenerative power generated during train braking has attracted a lot of attention lately. To utilize regenerative power with balancing the electric power supply-demand, we introduce a storage battery, and propose a charge control method of it. Our proposed algorithm could make not only balance the electric power supply-demand but also suppresses the fluctuation of the charged amount within the storage battery. The smaller amount of charge fluctuation, the smaller capacity battery would be available to use. In several existing methods, the empirical rules have been adopted to secure the balance, without consideration for suppressing the fluctuations of charged amount electricity. However, rule-based control which is based on the human empirical knowledge, has some limitations in electricity supply-demand dynamics in the railway systems. To overcome the limitations, we introduce reinforcement learning with an actor-critic algorithm to acquire the effective control policy which had been difficult to draw from the experts' knowledge as the rules. Through several computational simulations, we verified that the performance of our proposed method shows superior to that of the existing one.","PeriodicalId":248901,"journal":{"name":"2018 IEEE International Conference on Agents (ICA)","volume":"16 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-04-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128109451","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2018-04-12DOI: 10.1109/AGENTS.2018.8459966
Sotaro Tsutsui, Naoki Fukuta
Giving human knowledge to learning agents is a good way to speed up the process of reinforcement learning for learning agents. To give human knowledge to learning agent efficiently, it is important to estimate whether or not agents need more knowledge. In this paper, we present our approach to realize efficient teaching on an application which can show the users the progress of learning in a video game.
{"title":"Efficient Teaching Support to Non-player Learning Agents on Multiplayer Games","authors":"Sotaro Tsutsui, Naoki Fukuta","doi":"10.1109/AGENTS.2018.8459966","DOIUrl":"https://doi.org/10.1109/AGENTS.2018.8459966","url":null,"abstract":"Giving human knowledge to learning agents is a good way to speed up the process of reinforcement learning for learning agents. To give human knowledge to learning agent efficiently, it is important to estimate whether or not agents need more knowledge. In this paper, we present our approach to realize efficient teaching on an application which can show the users the progress of learning in a video game.","PeriodicalId":248901,"journal":{"name":"2018 IEEE International Conference on Agents (ICA)","volume":"49 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-04-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115038626","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2018-04-12DOI: 10.1109/AGENTS.2018.8460069
Tomohiro Nishida, Takanori Ito, Takayuki Ito
We have developed the consensus building support system called COLLAGREE on the Web with facilitator support function. If discussions are conduct by free participation on the Web, celebrity may participate. Therefore, in this work, we conduct large scale social experiments using support systems where celebrities participate. The purpose of this work is to verify the effectiveness and problems of consensus building support system in discussion where celebrities participate. IN the large-scale social experiments, we clarified that new threads by celebrities' increases the number of reply and approval number by about 3 times compared to other participants. Also, we clarified that opinion by celebrities did not little affect voting results. Furthermore, we clarified that participants were not influenced by celebrity opinions, but were influenced by other participants.
{"title":"Verification of Consensus Building Support System on Large-Scale Social Experiment where Celebrities Participate in Discussion","authors":"Tomohiro Nishida, Takanori Ito, Takayuki Ito","doi":"10.1109/AGENTS.2018.8460069","DOIUrl":"https://doi.org/10.1109/AGENTS.2018.8460069","url":null,"abstract":"We have developed the consensus building support system called COLLAGREE on the Web with facilitator support function. If discussions are conduct by free participation on the Web, celebrity may participate. Therefore, in this work, we conduct large scale social experiments using support systems where celebrities participate. The purpose of this work is to verify the effectiveness and problems of consensus building support system in discussion where celebrities participate. IN the large-scale social experiments, we clarified that new threads by celebrities' increases the number of reply and approval number by about 3 times compared to other participants. Also, we clarified that opinion by celebrities did not little affect voting results. Furthermore, we clarified that participants were not influenced by celebrity opinions, but were influenced by other participants.","PeriodicalId":248901,"journal":{"name":"2018 IEEE International Conference on Agents (ICA)","volume":"7 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-04-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133342199","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：