Search

38th European Photovoltaic Solar Energy Conference and Exhibition; 1668-1670, This paper proposes a novel electricity price prediction method using past electricity demand. The previous method predicts the electricity price using LSTM after wavelet transform of the past electricity price series. Wavelet transform decomposes the series into more smooth and stable series, and accurate prediction is performed by LSTM with these series. The proposed method also uses the electricity demand series, and the demand is also applied wavelet transform for effective LSTM step. Experimental results show the effectiveness of the proposed method, and the accuracy is drastically improved over the previous method.

36th European Photovoltaic Solar Energy Conference and Exhibition; 1877-1880, Solar energy is promising renewable energy source for sustainable society. It is important to fill in the temporal and spatial gaps to utilize the solar energy effectively. This paper proposes novel framework to realize spatiotemporal shifting of solar energy using EVs. Since EV is equipped with very large battery for long cruise distance, EV can be realize as energy carrier along with carrying people and luggage. Proposed framework indirectly controls the EV's battery charge/discharge based on power trade via EV aggregator, and the spatiotemporal shifting of solar energy is realized in the end. Experimental results show the effectiveness of proposed framework.

36th European Photovoltaic Solar Energy Conference and Exhibition; 1681-1684, Effective utilization of renewable energy is a big motivation to renovate the conventional power systems toward smart energy systems. However, renewable energy is often small scale and intermittent. Hence, the online energy management is necessary to balance between demand and energy production. This paper proposes an online multi-scale optimization framework for smart PV systems in order to reduce the inherent complexity when trying solve the complete problem at a fine-grain time resolution. The proposed framework realizes the effective energy management by solving several optimization problems covering multi time scale. Experimental results shows that the proposed framework can reduce the purchased energy from the utility grid by 19.1% compared with other baseline methods.