《Tab.1 Prediction errors of different models (in percentage)》

  提示：宽带有限、当前游客访问压缩模式

本系列图表出处文件名：随高清版一同展现

《考虑复杂关联关系深度挖掘的变压器状态参量预测方法》

1. 获取 高清版本忘记账户？点击这里登录

1. 下载图表忘记账户？点击这里登录

According to cases study，the GLSTM network can significantly improve the prediction accuracy，and the maximum prediction error is reduced from 15%to less than 10%compared with the single sequence prediction method without parameter correlation.Compared with the traditional AR，SVR，RBFNN and multi-parameter GM models，the error of GLSTM prediction model is greatly reduced，the error fluctuation range is smaller，and the prediction result is more stable.