查询结果:   李洁,林永峰.基于多时间尺度RNN的时序数据预测[J].计算机应用与软件,2018,35(7):33 - 37,62.
中文标题
基于多时间尺度RNN的时序数据预测
发表栏目
软件技术与研究
摘要点击数
325
英文标题
PREDICTION OF TIME SERIES DATA BASED ON MULTI-TIME SCALE RNN
作 者
李洁 林永峰 Li Jie Lin Yongfeng
作者单位
国网天津市电力公司电力科学研究院 天津 300380     
英文单位
Electric Power Research Institute, State Grid Tianjin Electric Power Company, Tianjin 300380,China     
关键词
时间序列数据 预测模型 后向传播算法 循环神经网络 多时间尺度
Keywords
Time series data Prediction model Backward propagation algorithm Recurrent neural network Multi-time scale
基金项目
天津市科技计划项目(17YFZCGX001610)
作者资料
李洁,工程师,主研领域:电网信息通信技术。林永峰,高级工程师。 。
文章摘要
时间序列数据的预测是很多领域研究的热点问题,但大多数模型都是在单一的时间尺度上进行研究。针对这一问题,基于真实的民航旅客历史出行记录,通过对数据统计处理,对旅客的出行特点以及行为规律进行了深入分析。根据其时序数据的特征建立基于后向传播算法的循环神经网络(RNN)预测模型,对未来时段的日客流量进行预测。在此基础上考虑到时序数据在不同时间尺度呈现不同的变化规律,建立多时间尺度的预测模型对旅客出行的周期性和趋势性进行建模,提升预测精度。
Abstract
The prediction of time series data is a hot topic in many fields, but most models are studied on a single time scale. In response to this problem, based on the real history of civil aviation travelers travel records, through the statistical processing of data, passenger travel characteristics and behavioral laws were analyzed in depth. Based on the characteristics of its time series data, a recurrent neural network (RNN) prediction model based on backward propagation algorithm was established to forecast the daily passenger traffic in the future. On the basis of this, taking into account that the time series data presents different patterns of change on different time scales, a multi-time scale prediction model was established to model the periodicity and trend of passenger travel and improved the prediction accuracy.
下载PDF全文