一个actor-critic算法,以最大限度地从直接甲醇燃料电池的电力交付

IF 49.7 1区 材料科学 Q1 ENERGY & FUELS
Hongbin Xu, Yang Jeong Park, Zhichu Ren, Daniel J. Zheng, Davide Menga, Haojun Jia, Chenru Duan, Guanzhou Zhu, Yuriy Román-Leshkov, Yang Shao-Horn, Ju Li
{"title":"一个actor-critic算法,以最大限度地从直接甲醇燃料电池的电力交付","authors":"Hongbin Xu, Yang Jeong Park, Zhichu Ren, Daniel J. Zheng, Davide Menga, Haojun Jia, Chenru Duan, Guanzhou Zhu, Yuriy Román-Leshkov, Yang Shao-Horn, Ju Li","doi":"10.1038/s41560-025-01804-x","DOIUrl":null,"url":null,"abstract":"<p>Optimizing nonlinear time-dependent control in complex energy systems such as direct methanol fuel cells (DMFCs) is a crucial engineering challenge. The long-term power delivery of DMFCs deteriorates as the electrocatalytic surfaces become fouled. Dynamic voltage adjustment can clean the surface and recover the activity of catalysts; however, manually identifying optimal control strategies considering multiple mechanisms is challenging. Here we demonstrated a nonlinear policy model (Alpha-Fuel-Cell) inspired by actor–critic reinforcement learning, which learns directly from real-world current–time trajectories to infer the state of catalysts during operation and generates a suitable action for the next timestep automatically. Moreover, the model can provide protocols to achieve the required power while significantly slowing the degradation of catalysts. Benefiting from this model, the time-averaged power delivered is 153% compared to constant potential operation for DMFCs over 12 hours. Our framework may be generalized to other energy device applications requiring long-time-horizon decision-making in the real world.</p>","PeriodicalId":19073,"journal":{"name":"Nature Energy","volume":"23 1","pages":""},"PeriodicalIF":49.7000,"publicationDate":"2025-07-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"An actor–critic algorithm to maximize the power delivered from direct methanol fuel cells\",\"authors\":\"Hongbin Xu, Yang Jeong Park, Zhichu Ren, Daniel J. Zheng, Davide Menga, Haojun Jia, Chenru Duan, Guanzhou Zhu, Yuriy Román-Leshkov, Yang Shao-Horn, Ju Li\",\"doi\":\"10.1038/s41560-025-01804-x\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"<p>Optimizing nonlinear time-dependent control in complex energy systems such as direct methanol fuel cells (DMFCs) is a crucial engineering challenge. The long-term power delivery of DMFCs deteriorates as the electrocatalytic surfaces become fouled. Dynamic voltage adjustment can clean the surface and recover the activity of catalysts; however, manually identifying optimal control strategies considering multiple mechanisms is challenging. Here we demonstrated a nonlinear policy model (Alpha-Fuel-Cell) inspired by actor–critic reinforcement learning, which learns directly from real-world current–time trajectories to infer the state of catalysts during operation and generates a suitable action for the next timestep automatically. Moreover, the model can provide protocols to achieve the required power while significantly slowing the degradation of catalysts. Benefiting from this model, the time-averaged power delivered is 153% compared to constant potential operation for DMFCs over 12 hours. Our framework may be generalized to other energy device applications requiring long-time-horizon decision-making in the real world.</p>\",\"PeriodicalId\":19073,\"journal\":{\"name\":\"Nature Energy\",\"volume\":\"23 1\",\"pages\":\"\"},\"PeriodicalIF\":49.7000,\"publicationDate\":\"2025-07-14\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Nature Energy\",\"FirstCategoryId\":\"88\",\"ListUrlMain\":\"https://doi.org/10.1038/s41560-025-01804-x\",\"RegionNum\":1,\"RegionCategory\":\"材料科学\",\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q1\",\"JCRName\":\"ENERGY & FUELS\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Nature Energy","FirstCategoryId":"88","ListUrlMain":"https://doi.org/10.1038/s41560-025-01804-x","RegionNum":1,"RegionCategory":"材料科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"ENERGY & FUELS","Score":null,"Total":0}
引用次数: 0

摘要

在复杂的能源系统中,如直接甲醇燃料电池(dmfc),优化非线性时变控制是一个关键的工程挑战。当电催化表面被污染时,dmfc的长期电力输送会恶化。动态电压调节可清洁表面,恢复催化剂活性;然而,人工识别考虑多种机制的最优控制策略是具有挑战性的。在这里,我们展示了一个非线性策略模型(Alpha-Fuel-Cell),该模型受到行动者-批评者强化学习的启发,它直接从现实世界的当前时间轨迹中学习,以推断催化剂在运行过程中的状态,并自动生成下一个时间步长的合适动作。此外,该模型可以提供实现所需功率的协议,同时显着减缓催化剂的降解。得益于该模型,与dmfc在12小时内恒电位运行相比,时间平均输出功率为153%。我们的框架可以推广到现实世界中需要长期决策的其他能源设备应用。
本文章由计算机程序翻译,如有差异,请以英文原文为准。

An actor–critic algorithm to maximize the power delivered from direct methanol fuel cells

An actor–critic algorithm to maximize the power delivered from direct methanol fuel cells

Optimizing nonlinear time-dependent control in complex energy systems such as direct methanol fuel cells (DMFCs) is a crucial engineering challenge. The long-term power delivery of DMFCs deteriorates as the electrocatalytic surfaces become fouled. Dynamic voltage adjustment can clean the surface and recover the activity of catalysts; however, manually identifying optimal control strategies considering multiple mechanisms is challenging. Here we demonstrated a nonlinear policy model (Alpha-Fuel-Cell) inspired by actor–critic reinforcement learning, which learns directly from real-world current–time trajectories to infer the state of catalysts during operation and generates a suitable action for the next timestep automatically. Moreover, the model can provide protocols to achieve the required power while significantly slowing the degradation of catalysts. Benefiting from this model, the time-averaged power delivered is 153% compared to constant potential operation for DMFCs over 12 hours. Our framework may be generalized to other energy device applications requiring long-time-horizon decision-making in the real world.

求助全文
通过发布文献求助,成功后即可免费获取论文全文。 去求助
来源期刊
Nature Energy
Nature Energy Energy-Energy Engineering and Power Technology
CiteScore
75.10
自引率
1.10%
发文量
193
期刊介绍: Nature Energy is a monthly, online-only journal committed to showcasing the most impactful research on energy, covering everything from its generation and distribution to the societal implications of energy technologies and policies. With a focus on exploring all facets of the ongoing energy discourse, Nature Energy delves into topics such as energy generation, storage, distribution, management, and the societal impacts of energy technologies and policies. Emphasizing studies that push the boundaries of knowledge and contribute to the development of next-generation solutions, the journal serves as a platform for the exchange of ideas among stakeholders at the forefront of the energy sector. Maintaining the hallmark standards of the Nature brand, Nature Energy boasts a dedicated team of professional editors, a rigorous peer-review process, meticulous copy-editing and production, rapid publication times, and editorial independence. In addition to original research articles, Nature Energy also publishes a range of content types, including Comments, Perspectives, Reviews, News & Views, Features, and Correspondence, covering a diverse array of disciplines relevant to the field of energy.
×
引用
GB/T 7714-2015
复制
MLA
复制
APA
复制
导出至
BibTeX EndNote RefMan NoteFirst NoteExpress
×
提示
您的信息不完整,为了账户安全,请先补充。
现在去补充
×
提示
您因"违规操作"
具体请查看互助需知
我知道了
×
提示
确定
请完成安全验证×
copy
已复制链接
快去分享给好友吧!
我知道了
右上角分享
点击右上角分享
0
联系我们:info@booksci.cn Book学术提供免费学术资源搜索服务,方便国内外学者检索中英文文献。致力于提供最便捷和优质的服务体验。 Copyright © 2023 布克学术 All rights reserved.
京ICP备2023020795号-1
ghs 京公网安备 11010802042870号
Book学术文献互助
Book学术文献互助群
群 号:604180095
Book学术官方微信