摘要
温度是一个涉及生活和生产的重要气象要素,为了更好地提高温度预报精度。论文基于LSTM神经网络提出了一种未来24小时温度滚动预报算法。实验研究表明,该方法在不同的各预报时段中非常稳定,且在未来24小时预报中表现较好,能够有效降低预报误差,提高预报精度,其预报结果平均绝对误差(MAE)为0.634,均方根误差(RMSE)为0.638,相关系数为0.999。
Temperature is an important meteorological element related to life and production,in order to better improve the ac-curacy of temperature prediction.Based on LSTM neural network,this paper proposes a rolling temperature prediction algorithm for the next 24 hours.The experimental research shows that this method is very stable in different forecast periods,and performs well in the future 24-hour forecast.It can effectively reduce the forecast error and improve the forecast accuracy.The mean absolute error(MAE)of the forecast results is 0.634,the root mean square error(RMSE)is 0.638,and the correlation coefficient is 0.999.
作者
雷鸣
陈凯华
郭阳
勾志竟
LEI Ming;CHEN Kaihua;GUO Yang;GOU Zhijing(Tianjin Meteorological Information Center,Tianjin 300074)
出处
《计算机与数字工程》
2025年第8期2112-2116,共5页
Computer & Digital Engineering
基金
国家自然科学基金项目(编号:41575156)
中央级公益性科研院所基本科研业务费专项(编号:IUMKY201605)
天津市气象局科研项目(编号:201914ybxm12)资助。
关键词
深度学习
LSTM
人工智能
温度预报
数据处理
deep learning
LSTM
artificial intelligence
temperature prediction
data processing
