现有深度强化学习(deep reinforcement learning,DRL)方法在解决配电网电压优化问题时,存在信用分配难、探索效率低等问题,在模型训练速度和优化效果等方面表现欠佳。为此,结合配电网分区降损与模仿学习的思想,提出一种基于指导信号的...现有深度强化学习(deep reinforcement learning,DRL)方法在解决配电网电压优化问题时,存在信用分配难、探索效率低等问题,在模型训练速度和优化效果等方面表现欠佳。为此,结合配电网分区降损与模仿学习的思想,提出一种基于指导信号的多智能体深度确定性策略梯度(guidance signal based multi-agent deep deterministic policy gradient,GS-MADDPG)的电压优化方法。首先,将电动汽车(electric vehicles,EV)集群、分布式电源(distributed generations,DG)和无功调节装置作为决策智能体,构建强化学习优化模型。然后,通过配电网分区,解耦多智能体的外部奖励,并结合模仿学习,利用指导信号引入内部奖励,帮助智能体快速寻优。最后,基于改进IEEE33节点系统进行算例测试。结果表明,所提电压优化策略较传统DRL方法具有更高的样本利用率,实现了更稳定的收敛及更高的模型训练效率,提升了配电网电压的优化效果。展开更多
AIM： To elucidate the sequential gene expression profile in AGS cells co-cultured with wild-type Helicobacter pylori （H pylon） as a model of Hpylori-infected gastric epithelium, and to further examine the contribut...AIM： To elucidate the sequential gene expression profile in AGS cells co-cultured with wild-type Helicobacter pylori （H pylon） as a model of Hpylori-infected gastric epithelium, and to further examine the contribution of cag-pathogenicity islands （cagPAI）-coding type IV secretion system and the two pathways, nuclear factor kappa B （NF-κB） and extracellular signal-regulated kinases （ERK） on wild-type Hpylori-induced gene expression. METHODS： Gene expression profiles induced by Hpylori were evaluated in AGS gastric epithelial cells using cDNA microarray, which were present in the 4600 independent clones picked up from the human gastric tissue. We also analyzed the contribution of NF-κB and ERK signaling on H pylori-induced gene expression by using inhibitors of specific signal pathways. The isogenic mutant with disrupted cagE （△cagE） was used to elucidate the role of cagPAI-encoding type IV secretion system in the gene expression profile. RESULTS： According to the expression profile, the genes were classified into four clusters. Among them, the clusters characterized by continuous upregulation were most conspicuous, and it contained many signal transducer activity-associated genes. The role of cagPAI on cultured cells was also investigated using isogenic mutant cagE, which carries non-functional cagPAI. Then the upregulation of more than 80% of the induced genes （476/566） was found to depend on cagPAI. Signal transducer pathway through NF-κB or ERK are the major pathways which are known to be activated by cagPAI-positive H pylori. The role of these pathways in the whole signal activation by cagPAI-positive H pyloriwas analyzed. The specific inhibitors against NF-κB or ERK pathway blocked the activation of gene expression in 65% （367/566） or 76% （429/566） of the genes whose activation appealed to depend on cagPAI. CONCLUSION： These results suggest that more than half of the genes induced by cayPAI-positive H pylori depend on NF-κB and ERK signaling activation, and these pathways may play a role in the gene expression induced by hostbacterial interaction which may associate with H pylorirelated gastro-duodenal diseases.展开更多
文摘现有深度强化学习(deep reinforcement learning,DRL)方法在解决配电网电压优化问题时,存在信用分配难、探索效率低等问题,在模型训练速度和优化效果等方面表现欠佳。为此,结合配电网分区降损与模仿学习的思想,提出一种基于指导信号的多智能体深度确定性策略梯度(guidance signal based multi-agent deep deterministic policy gradient,GS-MADDPG)的电压优化方法。首先,将电动汽车(electric vehicles,EV)集群、分布式电源(distributed generations,DG)和无功调节装置作为决策智能体,构建强化学习优化模型。然后,通过配电网分区,解耦多智能体的外部奖励,并结合模仿学习,利用指导信号引入内部奖励,帮助智能体快速寻优。最后,基于改进IEEE33节点系统进行算例测试。结果表明,所提电压优化策略较传统DRL方法具有更高的样本利用率,实现了更稳定的收敛及更高的模型训练效率,提升了配电网电压的优化效果。
基金Supported by a Grant-in-Aid for Scientific Research from the Ministry of Education, Science, Sport, and Culture of Japan
文摘AIM： To elucidate the sequential gene expression profile in AGS cells co-cultured with wild-type Helicobacter pylori （H pylon） as a model of Hpylori-infected gastric epithelium, and to further examine the contribution of cag-pathogenicity islands （cagPAI）-coding type IV secretion system and the two pathways, nuclear factor kappa B （NF-κB） and extracellular signal-regulated kinases （ERK） on wild-type Hpylori-induced gene expression. METHODS： Gene expression profiles induced by Hpylori were evaluated in AGS gastric epithelial cells using cDNA microarray, which were present in the 4600 independent clones picked up from the human gastric tissue. We also analyzed the contribution of NF-κB and ERK signaling on H pylori-induced gene expression by using inhibitors of specific signal pathways. The isogenic mutant with disrupted cagE （△cagE） was used to elucidate the role of cagPAI-encoding type IV secretion system in the gene expression profile. RESULTS： According to the expression profile, the genes were classified into four clusters. Among them, the clusters characterized by continuous upregulation were most conspicuous, and it contained many signal transducer activity-associated genes. The role of cagPAI on cultured cells was also investigated using isogenic mutant cagE, which carries non-functional cagPAI. Then the upregulation of more than 80% of the induced genes （476/566） was found to depend on cagPAI. Signal transducer pathway through NF-κB or ERK are the major pathways which are known to be activated by cagPAI-positive H pylori. The role of these pathways in the whole signal activation by cagPAI-positive H pyloriwas analyzed. The specific inhibitors against NF-κB or ERK pathway blocked the activation of gene expression in 65% （367/566） or 76% （429/566） of the genes whose activation appealed to depend on cagPAI. CONCLUSION： These results suggest that more than half of the genes induced by cayPAI-positive H pylori depend on NF-κB and ERK signaling activation, and these pathways may play a role in the gene expression induced by hostbacterial interaction which may associate with H pylorirelated gastro-duodenal diseases.
