The key challenge for household load forecasting lies in the high volatility and uncertainty of load profiles. Traditional methods tend to avoid such uncertainty by load aggregation (to offset uncertainties), customer classification (to cluster uncertainties) and spectral analysis (to filter out uncertainties). This paper, for the first time, aims to directly learn the uncertainty by applying a new breed of machine learning algorithms -deep learning. However simply adding layers in neural networks will cap the forecasting performance due to the occurrence of overfitting. A novel pooling-based deep recurrent neural network (PDRNN) is proposed in this paper which batches a group of customers' load profiles into a pool of inputs. Essentially the model could address the over-fitting issue by increasing data diversity and volume. This work reports the first attempts to develop a bespoke deep learning application for household load forecasting and achieved preliminary success. The developed method is implemented on Tensorflow deep learning platform and tested on 920 smart metered customers from Ireland. Compared with the state-of-art techniques in household load forecasting, the proposed method outperforms ARIMA by 19.5%, SVR by 13.1% and classical deep RNN by 6.5% in terms of RMSE.

No abstract

No abstract

Phase imbalance in the UK and European low voltage (415V, LV) distribution networks causes additional energy losses.A key barrier against understanding the imbalanceinduced energy losses is the absence of high-resolution time-series data for LV networks. It remains a challenge to estimate imbalance-induced energy losses in LV networks that only have the yearly average currents of the three phases. To address this insufficient data challenge, this paper proposes a new customized statistical approach, named as the CCRE (Clustering, Classification, and Range Estimation) approach. It finds a match between the network with only the yearly average phase currents (the data-scarce network) and a cluster of networks with time series of phase current data (data-rich networks). Then CCRE performs a range estimation of the imbalance-induced energy loss for the cluster of data-rich networks that resemble the data-scarce network. The Chebyshev's inequality is applied to narrow down this range, which represents the confidence interval of the imbalance-induced energy loss for the data-scarce network. Case studies reveal that, given such few data from the data-scarce networks, more than 80% of these networks are classified to the correct clusters and the confidence of the imbalance-induced energy loss estimation is 89%.

Adult hippocampal neurogenesis occurs in the dentate gyrus (DG) of the mouse hippocampus, and plays roles in learning and memory progresses. In amyloid precursor protein (APP)/presenilin 1 (PS1) transgenic mice, a rodent model of Alzheimer's disease (AD), severe impairment of neurogenesis in the dentate subgranular zone (SGZ) of the DG has been reported. Osthole, an active constituent of Cnidium monnieri (L.) CUSSON, has been reported to exert neuroprotective effects and may promote neural stem cell proliferation. However, whether osthole ameliorates spatial memory deficits and improves hippocampal neurogenesis in APP/PS1 mice remains unknown. In this study we found that osthole (30 mg/kg intraperitoneally (i.p.) once daily) treatment dramatically ameliorated the cognitive impairments by Morris Water Maze test and passive avoidance test, and augmented neurogenesis in the DG of hippocampus in APP/PS1 mice. Furthermore, osthole treatment upregulated expression of brain-derived neurotrophic factor (BDNF) and enhanced activation of the BDNF receptor tyrosine receptor kinase B (TrkB) following increased phosphorylation of cyclic AMP response element-binding protein (CREB), indicating that osthole improves neurogenesis via stimulating BDNF/TrkB/CREB signaling in APP/PS1 transgenic mice.