场景生成技术通过模拟多种典型场景的随机性和多样性,为解决历史数据样本有限和极端场景难以覆盖的问题提供了重要数据支撑。该文提出一种基于特征约束与目标驱动的WGAN-GP(Wasserstein GAN with gradient penalty)光伏多场景生成方法...场景生成技术通过模拟多种典型场景的随机性和多样性,为解决历史数据样本有限和极端场景难以覆盖的问题提供了重要数据支撑。该文提出一种基于特征约束与目标驱动的WGAN-GP(Wasserstein GAN with gradient penalty)光伏多场景生成方法。该方法通过引入Wasserstein距离及梯度惩罚,提高了生成器的稳定性与生成样本的多样性;在此基础上,提取晴天、阴雨天和多云天3类典型场景的关键特征,明确生成目标,并在生成器的输出阶段嵌入动态门函数,动态划定白天与夜间的分界点,确保生成的夜间输出为零,白天的辐照度变化符合实际物理规律;此外,采用加权采样优化策略,通过动态调整样本权重和选中概率,进一步强化对关键特性样本的学习,使生成器能够更准确地捕捉目标特性,从而显著提升稀缺场景的生成效果。算例结果表明,该方法能够精准捕捉不同天气场景的关键特征,生成样本在目标特性匹配及物理合理性方面表现良好,为光伏场景生成提供了一种可靠的解决方案。展开更多
Fast and accurate transient stability analysis is crucial to power system operation.With high penetration level of wind power resources,practical dynamic security region(PDSR)with hyper plane expression has outstandin...Fast and accurate transient stability analysis is crucial to power system operation.With high penetration level of wind power resources,practical dynamic security region(PDSR)with hyper plane expression has outstanding advantages in situational awareness and series of optimization problems.The precondition of obtaining accurate PDSR boundary is to locate sufficient points around the boundary(critical points).Therefore,this paper proposes a space division and Wasserstein generative adversarial network with gra-dient penalty(WGAN-GP)based fast generation method of PDSR boundary.First,the typical differential topological characterizations of dynamic security region(DSR)provide strong theoretical foundation that the interior of DSR is hole-free and the boundaries of DSR are tight and knot-free.Then,the space division method is proposed to calculate critical operation area where the PDSR boundary is located,tremen-dously compressing the search space to locate critical points and improving the confidence level of boundary fitting result.Furthermore,the WGAN-GP model is utilized to fast obtain large number of criti-cal points based on learning the data distribution of the small training set aforementioned.Finally,the PDSR boundary with hyperplanes is fitted by the least square method.The case study is tested on the Institute of Electrical and Electronics Engineers(IEEE)39-bus system and the results verify the accuracy and efficiency of the proposed method.展开更多
针对水稻病害图像数据集样本较少而影响深度神经网络模型学习的精度问题,提出一种改进的对抗生成网络模型ViT-WGAN-GP(Vision Transformer and Wasserstein Generative Adversarial Networks with Gradient Penalty)用于对图像数据集进...针对水稻病害图像数据集样本较少而影响深度神经网络模型学习的精度问题,提出一种改进的对抗生成网络模型ViT-WGAN-GP(Vision Transformer and Wasserstein Generative Adversarial Networks with Gradient Penalty)用于对图像数据集进行增强。首先在生成模型引入Vision Transformer结构加强对全局特征的学习;其次在判别模型采用WGAN-GP结构,通过Wasserstein衡量函数和梯度惩罚项保证模型训练的稳定性,提升生成图像的效果;最后使用增强后的样本集训练深度神经网络模型。实验结果表明,针对水稻病害图像,ViT-WGAN-GP模型与GAN、WGAN-GP相比生成图像效果提升显著。使用增强后的水稻病害样本集训练VGG16、ResNet34和GoogLeNet模型,水稻病害识别平均准确率分别达到94.3%,96.2%,97.5%,分别提升了9.7%,2.8%,4.8%。由此可见,该ViT-WGAN-GP模型能生成较为真实的水稻病害图像,且能在小样本集下,较大幅度提高深度神经网络模型的识别准确率。展开更多
In order to improve the quality of low-dose computational tomography (CT)images, the paper proposes an improved image denoising approach based on WGAN-gpwith Wasserstein distance. For improving the training and the co...In order to improve the quality of low-dose computational tomography (CT)images, the paper proposes an improved image denoising approach based on WGAN-gpwith Wasserstein distance. For improving the training and the convergence efficiency, thegiven method introduces the gradient penalty term to WGAN network. The novelperceptual loss is introduced to make the texture information of the low-dose imagessensitive to the diagnostician eye. The experimental results show that compared with thestate-of-art methods, the time complexity is reduced, and the visual quality of low-doseCT images is significantly improved.展开更多
本文提出了一种基于改进Wasserstein生成式对抗网络(De-aliasing Wasserstein Generative Adversarial Network with Gradient Penalty,DAWGAN-GP)的磁共振图像重构算法,该方法利用Wasserstein生成式对抗网络代替传统的生成式对抗网络,...本文提出了一种基于改进Wasserstein生成式对抗网络(De-aliasing Wasserstein Generative Adversarial Network with Gradient Penalty,DAWGAN-GP)的磁共振图像重构算法,该方法利用Wasserstein生成式对抗网络代替传统的生成式对抗网络,并结合梯度惩罚的方法提高训练速度,解决WGAN收敛缓慢问题.此外,为了有更好的重构效果,我们将感知损失,像素损失和频域损失引入至损失函数中进行网络训练.实验结果表明,对比现有的基于深度学习的磁共振图像重构算法,基于DAWGAN-GP的磁共振图像重构方法具有更好的重构效果,可获得更高的峰值信噪比(Peak Signal to Noise Ratio,PSNR)和更好的结构相似性(Structural Similarity Index Measure,SSIM).展开更多
文摘场景生成技术通过模拟多种典型场景的随机性和多样性,为解决历史数据样本有限和极端场景难以覆盖的问题提供了重要数据支撑。该文提出一种基于特征约束与目标驱动的WGAN-GP(Wasserstein GAN with gradient penalty)光伏多场景生成方法。该方法通过引入Wasserstein距离及梯度惩罚,提高了生成器的稳定性与生成样本的多样性;在此基础上,提取晴天、阴雨天和多云天3类典型场景的关键特征,明确生成目标,并在生成器的输出阶段嵌入动态门函数,动态划定白天与夜间的分界点,确保生成的夜间输出为零,白天的辐照度变化符合实际物理规律;此外,采用加权采样优化策略,通过动态调整样本权重和选中概率,进一步强化对关键特性样本的学习,使生成器能够更准确地捕捉目标特性,从而显著提升稀缺场景的生成效果。算例结果表明,该方法能够精准捕捉不同天气场景的关键特征,生成样本在目标特性匹配及物理合理性方面表现良好,为光伏场景生成提供了一种可靠的解决方案。
基金funded in part by the National Key Research and Development Program of China(2020YFB0905900)in part by Science and Technology Project of State Grid Corporation of China(SGCC)The Key Technologies for Electric Internet of Things(SGTJDK00DWJS2100223).
文摘Fast and accurate transient stability analysis is crucial to power system operation.With high penetration level of wind power resources,practical dynamic security region(PDSR)with hyper plane expression has outstanding advantages in situational awareness and series of optimization problems.The precondition of obtaining accurate PDSR boundary is to locate sufficient points around the boundary(critical points).Therefore,this paper proposes a space division and Wasserstein generative adversarial network with gra-dient penalty(WGAN-GP)based fast generation method of PDSR boundary.First,the typical differential topological characterizations of dynamic security region(DSR)provide strong theoretical foundation that the interior of DSR is hole-free and the boundaries of DSR are tight and knot-free.Then,the space division method is proposed to calculate critical operation area where the PDSR boundary is located,tremen-dously compressing the search space to locate critical points and improving the confidence level of boundary fitting result.Furthermore,the WGAN-GP model is utilized to fast obtain large number of criti-cal points based on learning the data distribution of the small training set aforementioned.Finally,the PDSR boundary with hyperplanes is fitted by the least square method.The case study is tested on the Institute of Electrical and Electronics Engineers(IEEE)39-bus system and the results verify the accuracy and efficiency of the proposed method.
基金supported by National Natural Science Foundation ofChina (61672279)Project of “Six Talents Peak” in Jiangsu (2012-WLW-023)OpenFoundation of State Key Laboratory of Hydrology-Water Resources and HydraulicEngineering, Nanjing Hydraulic Research Institute, China (2016491411).
文摘In order to improve the quality of low-dose computational tomography (CT)images, the paper proposes an improved image denoising approach based on WGAN-gpwith Wasserstein distance. For improving the training and the convergence efficiency, thegiven method introduces the gradient penalty term to WGAN network. The novelperceptual loss is introduced to make the texture information of the low-dose imagessensitive to the diagnostician eye. The experimental results show that compared with thestate-of-art methods, the time complexity is reduced, and the visual quality of low-doseCT images is significantly improved.
文摘本文提出了一种基于改进Wasserstein生成式对抗网络(De-aliasing Wasserstein Generative Adversarial Network with Gradient Penalty,DAWGAN-GP)的磁共振图像重构算法,该方法利用Wasserstein生成式对抗网络代替传统的生成式对抗网络,并结合梯度惩罚的方法提高训练速度,解决WGAN收敛缓慢问题.此外,为了有更好的重构效果,我们将感知损失,像素损失和频域损失引入至损失函数中进行网络训练.实验结果表明,对比现有的基于深度学习的磁共振图像重构算法,基于DAWGAN-GP的磁共振图像重构方法具有更好的重构效果,可获得更高的峰值信噪比(Peak Signal to Noise Ratio,PSNR)和更好的结构相似性(Structural Similarity Index Measure,SSIM).
