The real-time AC optimal power flow(OPF)problem is a key issue in making fast and accurate decisions to ensure the safety and economy of power systems.With the rapid development of renewable energies,the fluctuation h...The real-time AC optimal power flow(OPF)problem is a key issue in making fast and accurate decisions to ensure the safety and economy of power systems.With the rapid development of renewable energies,the fluctuation has grown more vibrant,thus a novel approach called safe deep reinforcement learning is proposed in this paper.Herein,the real-time ACOPF problem is modeled as a constrained Markov decision process,and primal-dual optimization(PDO)based proximal policy optimization(PPO)is used to learn the optimal generator outputs in the primal domain and security constraints in the dual domain,which avoids manually selecting a trade-off between penalties for constraint violations and rewards for the economy.Before training,behavior cloning clones the expert experience into the initial weights of neural networks.Moreover,multiprocessing training is utilized to accelerate the training speed.Case studies are conducted on the IEEE 118-bus system and the modified IEEE 118-bus system.Compared with other methods,the experimental results show that the proposed method can achieve security and near-optimal economic goals by fast calculating the real-time ACOPF problem.展开更多
The energy industry,now in an era of digitization driven by computational design,is gradually moving towards automating the entire process from computational prediction to device assembly,aiming to minimize the relian...The energy industry,now in an era of digitization driven by computational design,is gradually moving towards automating the entire process from computational prediction to device assembly,aiming to minimize the reliance on time-consuming,manual trial-and-error validation.In this study,guided by computational density functional theory(DFT)predictions,a humanoid robotic arm,based on artificial intelligence(AI),was creatively utilized to assemble clean energy devices,solid oxide fuel cells(SOFCs).The material La_(0.35)Bi_(0.15)Sr_(0.5)FeO_(3-δ)(LBSF)was DFT-predicted to have high oxygen reduction reactions(ORRs)ability,suitable for the cathode in SOFCs compared to the conventional La_(0.5)Sr_(0.5)FeO_(3-δ)(LSF).The material was made into ink then passed to the assembly platform with AI-driven robotics.AI-driven robotics was employed with an imitation learning method to effectively learn skills directly from human demonstrations,thereby alleviating researchers from labor-intensive tasks.We demonstrate our approach for autonomous SOFCs fabrication.For easy platform usage in the future,Large Language Models(LLMs)were incorporated to understand human commands.Visual information was captured by an RGBD camera to identify and locate the cathode painting spot.An imitation learning framework was then applied to learn the painting path from human operations and can be generalized to different conditions.The auto-fabricated single cells with the DFT-predicted LBSF cathode were tested and achieved a power density of 966mW∕cm^(2)at 700℃,more than double the performance of LSF.By integrating computational design with an AI-driven assembly platform,this study marks an initial step towards an AI-driven material lab,exponentially accelerating material design in the near future.The platform can also help disabled researchers achieve their ideas through the behavior cloning approach.展开更多
基金supported by the National Natural Science Foundation of China(52007173 and U22B2098).
文摘The real-time AC optimal power flow(OPF)problem is a key issue in making fast and accurate decisions to ensure the safety and economy of power systems.With the rapid development of renewable energies,the fluctuation has grown more vibrant,thus a novel approach called safe deep reinforcement learning is proposed in this paper.Herein,the real-time ACOPF problem is modeled as a constrained Markov decision process,and primal-dual optimization(PDO)based proximal policy optimization(PPO)is used to learn the optimal generator outputs in the primal domain and security constraints in the dual domain,which avoids manually selecting a trade-off between penalties for constraint violations and rewards for the economy.Before training,behavior cloning clones the expert experience into the initial weights of neural networks.Moreover,multiprocessing training is utilized to accelerate the training speed.Case studies are conducted on the IEEE 118-bus system and the modified IEEE 118-bus system.Compared with other methods,the experimental results show that the proposed method can achieve security and near-optimal economic goals by fast calculating the real-time ACOPF problem.
文摘The energy industry,now in an era of digitization driven by computational design,is gradually moving towards automating the entire process from computational prediction to device assembly,aiming to minimize the reliance on time-consuming,manual trial-and-error validation.In this study,guided by computational density functional theory(DFT)predictions,a humanoid robotic arm,based on artificial intelligence(AI),was creatively utilized to assemble clean energy devices,solid oxide fuel cells(SOFCs).The material La_(0.35)Bi_(0.15)Sr_(0.5)FeO_(3-δ)(LBSF)was DFT-predicted to have high oxygen reduction reactions(ORRs)ability,suitable for the cathode in SOFCs compared to the conventional La_(0.5)Sr_(0.5)FeO_(3-δ)(LSF).The material was made into ink then passed to the assembly platform with AI-driven robotics.AI-driven robotics was employed with an imitation learning method to effectively learn skills directly from human demonstrations,thereby alleviating researchers from labor-intensive tasks.We demonstrate our approach for autonomous SOFCs fabrication.For easy platform usage in the future,Large Language Models(LLMs)were incorporated to understand human commands.Visual information was captured by an RGBD camera to identify and locate the cathode painting spot.An imitation learning framework was then applied to learn the painting path from human operations and can be generalized to different conditions.The auto-fabricated single cells with the DFT-predicted LBSF cathode were tested and achieved a power density of 966mW∕cm^(2)at 700℃,more than double the performance of LSF.By integrating computational design with an AI-driven assembly platform,this study marks an initial step towards an AI-driven material lab,exponentially accelerating material design in the near future.The platform can also help disabled researchers achieve their ideas through the behavior cloning approach.
