• 主题
高级搜索  >
历史搜索 清空历史
首页> 外文期刊>IEEJ energy journal >Short-term Load Forecasting Using Artificial Neural Network -Assessment for the 10 Areas in Japan-
【24h】

Short-term Load Forecasting Using Artificial Neural Network -Assessment for the 10 Areas in Japan-

机译：基于人工神经网络的短期负荷预测-日本10个地区的评估-

获取原文
获取原文并翻译 | 示例
           
页面导航
  • 摘要
  • 著录项
  • 相似文献
  • 相关主题

摘要

For this paper, we attempted to forecast next-day electricity load in the 10 Japanese areas by using artificial neural networks, a type of artificial intelligence technology. The model used here conducts a principal component analysis of daily load curves and utilizes selective ensembling, indicating good forecasting performance with the mean absolute percentage error limited to less than 2.5%. However, forecasting performances differ widely from season to season, with the percentage error ranging wide from 1.4% in Tohoku to 2.7% in Chugoku and Kyushu. Seasonal changes in errors in Hokkaido and Okinawa differ from those in other regions, reflecting regional characteristics. As the model used here is a simple one using meteorological data at only one location in each area, more massive data may be used to further improve forecasting performance. Even if forecasting performance is improved in a manner to extend the model, however, summer and winter forecasting errors may still be large. It would be a key future research challenge to consider an advanced forecasting technique assuming abnormal weather conditions and weather forecasting errors.
机译：在本文中，我们尝试通过使用一种人工智能技术的人工神经网络来预测日本10个地区的次日用电负荷。这里使用的模型对日负荷曲线进行主成分分析，并利用选择性汇总，表明了良好的预测性能，平均绝对百分比误差限制为小于2.5％。但是，各个季节的预测表现差异很大，百分比误差范围从东北的1.4％到中国和九州的2.7％不等。北海道和冲绳的错误季节性变化与其他区域不同，反映了区域特征。由于此处使用的模型是一种简单的模型，仅在每个区域的一个位置使用气象数据，因此可以使用更大量的数据来进一步提高预测性能。但是，即使以扩展模型的方式提高了预测性能，夏季和冬季的预测误差仍然可能很大。考虑到假设异常天气情况和天气预报误差的先进预报技术，这将是未来研究的关键挑战。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱：kefu@zhangqiaokeyan.com

京公网安备：11010802029741号 ICP备案号：京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有

  • 客服微信

  • 服务号