Search

With the increase in power system load, the number of switch cabinets increases rapidly, and the accidents caused by DC operating circuits become more and more obvious. Abnormal detection has become an important part of the DC operating circuit. Abnormal data, mainly voltage measurement, often contain important information. Finding and analyzing these abnormal data can give early warning for circuit failure or failure. However, abnormal data is often difficult to obtain and lacks marks, that is, the problem of sample imbalance and unsupervised problems, resulting in abnormal detection methods that can not meet the increasing accuracy and efficiency requirements. Generative adversarial network (GAN) is an unsupervised deep learning model, which has been successfully applied to unsupervised time series anomaly detection. However, GAN has technical difficulties in the anomaly detection process, and there is still room for optimization in accuracy. In this paper, the unsupervised time series anomaly detection method based on GAN is studied, and an anomaly detection method of a DC operating circuit combined with LSTM and GAN is designed.

Serum prealbumin is a recognized marker of malnutrition, but its prognostic role in patients with hemorrhagic stroke remains unclear. In this study, we retrospectively reviewed the records of 105 patients with hemorrhagic stroke admitted to Renmin Hospital of Wuhan University, China, from January to December 2015. We collected demographic and radiological data, and recorded serum prealbumin levels at admission and on days 1, 3, 6, 9, and 14-21. The existence of infections and gastrointestinal hemorrhage, and clinical condition at discharge were also recorded. Serum prealbumin levels during hospitalization were significantly lower in patients with infections compared with those without infections, and also significantly lower in patients with gastrointestinal hemorrhage compared with those without. Serum prealbumin levels at discharge were significantly higher in patients with good recovery than in those with poor recovery. We conclude that regular serum prealbumin measurements in patients with hemorrhagic stroke may be a useful indicator for determining clinical status and prognosis, which may therefore help to guide clinical decision-making.

The radiotherapy as a local and regional modality is widely applied in treatment of glioma, but most glioblastomas are commonly resistant to irradiation treatment. It remains challengeable to seek out efficient strategies to conquer the resistance of human glioblastoma cells to radiotherapy. Leucine-rich repeats and immunoglobulin-like domains protein 1 (LRIG1) is a newly discovered tumor suppressor which involved in regulation of chemosensitivity in various human cancer cells. In the present study, we established a radioresistant U251 cell line (U251R) to investigate the role of LRIG1 in regulation of radiosensitivity in human glioblastoma cells. Significantly decreased expression level of LRIG1 and enhanced expression of EGFR and phosphorylated Akt were detected in U251R cells compared with the parental U251 cells. U251R cells exhibited an advantage in colony formation ability, which accompanied by remarkably reduced X-ray-induced γ-H2AX foci formation and cell apoptosis. LRIG1 overexpression significantly inhibited the colony formation ability of U251R cells and obviously enhanced X-ray-inducedγ-H2AX foci formation and cell apoptosis. In addition, up-regulated expression of LRIG1 suppressed the expression level of EGFR and phosphorylated Akt protein. Our results demonstrated that LRIG1 expression was related to the radiosensitivity of human glioblastoma cells and may play an important role in the regulation of cellular radiosensitivity of human glioblastoma cells through the EGFR/Akt signaling pathway.