Gaoya dianqi油中溶解乙炔作为电力变压器中重要的放电程度表征参量之一，对其进行多步预测可以为变压器故障诊断及预警提供重要依据。现有的状态预测模型主要集中于单步预测，对于未来更长时期变化趋势的预测手段不足。此外，基于深度学习的多步预测模型的超参数选择大多基于经验和朴素的单一控制变量法，超参数之间的耦合关系没有得到充分的研究。文中提出基于多输出策略的门控循环单元（gated recurrent unit，GRU）神经网络多步预测模型，通过改变模型结构超参数和训练超参数研究超参数之间的耦合关系 ...
赵军 +6 more
doaj
[Optical coherence tomography angiography image segmentation based on multi-scale dilated convolution and dual attention network]. [PDF]
Sheng Wu Yi Xue Gong Cheng Xue Za ZhiZeng Q, Xiao Z, He B, Peng R.
europepmc +1 more source
[Cluster-guided adaptive Transformer for muscle fatigue prediction]. [PDF]
Sheng Wu Yi Xue Gong Cheng Xue Za ZhiFan B, Bao X, Ding L, Wu J.
europepmc +1 more source
[Collaborative motion control method for dual-arm surgical robots based on hybrid multi-stage optimization]. [PDF]
Sheng Wu Yi Xue Gong Cheng Xue Za ZhiLin B, Li K.
europepmc +1 more source
Gaoya dianqi为解决配电网中部分配电变压器在长期运行后出现的导线接头老化，局部温升严重等安全隐患发现不及时的问题，建立了针对台区配电变压器三相四线制接触阻抗高准确性计算模型，形成了数据驱动的配电变压器潜在安全隐患温升预警方法。利用三相四线制台区变压器的等值电路模型，构建了计算接触阻抗的等价方程组。因方程组变量数大于方程组数目，特将不可解方程组模型转化为求最小值的等效优化模型，并运用BFGS优化算法对模型进行优化迭代求解。利用最大负载电流下的长期发热模型，建立了接触阻抗与温度的等价关系，形成了接触阻抗的分级预警体系 ...
陈春 +4 more
doaj
[Prior knowledge-guided and border-focused segmentation of ischemic stroke lesions]. [PDF]
Sheng Wu Yi Xue Gong Cheng Xue Za ZhiCui W, Wen Z, Shao H.
europepmc +1 more source
[Recent advances in individualized treatment for pediatric high-risk B-cell acute lymphoblastic leukemia]. [PDF]
Zhongguo Dang Dai Er Ke Za ZhiLi YQ, Wang YJ, Li ZZ, Ni JX.
europepmc +1 more source
Gaoya dianqi变压器是电力供应的重要组成部分，油中溶解气体测试方法通常用于配电变压器的质量检测，从而诊断其故障。文中提出了一种改进的优化算法，利用溶解气体分析（disslved gases analysis，DGA）对电力变压器质量检测从而进行故障预测。首先，从各种来源收集来自变压器DGA的数据，并选择最佳混合特征集作为模型的输入。其次，对于特征选择，采用核主成分分析（kernel principal component analysis，KPCA），为了优化最小二乘支持向量机（least squares ...
田霖 +4 more
doaj
[Development of a machine learning-based ionization efficiency prediction model for per- and polyfluoroalkyl substances and its application in semi-quantitative analysis]. [PDF]
Se PuSun SZ +6 more
europepmc +1 more source
[Swin-ResViT network for real-time generation of high-quality localization MR images from low-quality cine-MR]. [PDF]
Nan Fang Yi Ke Da Xue Xue BaoChen B, Wang X, Zhao X, Song T, Li Y.
europepmc +1 more source