Load forecasting method for Commercial facilities by determination of working time and considering weather information
2018 7th International Conference on Renewable Energy Research and Applications (ICRERA)(2018)
摘要
In recent years, spread of BEMS (Building Energy Management System) is expected for the effective use of renewable energy and saving energy. On the other hand, we need energy management because renewable energy is affected by weather. By forecasting load curves of the next day based on the tendency of past loads, we can manage storage battery operation and demand response. However, in case of load forecasting, two factors should be considered. The first is that the working hours differs in each commercial facility. The second is that the maximum load is different by weather condition and the day of week. In this paper, we propose a load forecasting method focused on the working hours and weather information. Firstly, daily load profiles were categorized into the working day and the non-working day for each commercial facility by using threshold calculated from daily load values of 2270 commercial facilities in Kanto area in Japan. Next, the working time of each facility was decided by analyzing past load values until forecasting date, and clustering was performed with load curves in working time. The cluster which forecasting date belonged to was decided by weather information of forecasting date, load values of the previous day and the day of week. Base load of target facility was calculated by using load curves which belong to correspond cluster. By using parameter with the strongest correlation coefficient between maximum load and weather information, we performed linear regression and calculated the maximum load of forecasting day. As a comparison model to evaluate the forecasting accuracy of the proposed method, a simple persistence model was created. The proposed model was superior to a persistence model.
更多查看译文
关键词
Load forecasting BEMS,Working time,Clustering,Commercial facilities
AI 理解论文
溯源树
样例
![](https://originalfileserver.aminer.cn/sys/aminer/pubs/mrt_preview.jpeg)
生成溯源树,研究论文发展脉络
Chat Paper
正在生成论文摘要