.png){kind=icon}
返回
Simultaneous multistep transformer architecture for model predictive control
.png){kind=icon}
PRE
AI
DOI：10.1016/j.compchemeng.2023.108396.png)
.png){kind=icon}
原文求助.png){kind=icon}
分享.png){kind=icon}
收藏
摘要
En 中文
Transformer neural networks have revolutionized natural language processing by effectively addressing the vanishing gradient problem. This study focuses on applying Transformer models to time-series forecasting and customizing them for a simultaneous multistep-ahead prediction model in surrogate model predictive control (MPC). The proposed method showcases improved control performance and computational efficiency compared to LSTM-based MPC and one-step-ahead prediction models using both LSTM and Transformer networks. The study introduces three key contributions: (1) a new MPC system based on a Transformer time-series architecture, (2) a training method enabling multistep-ahead prediction for time-series machine learning models, and (3) validation of the enhanced time performance of multistep-ahead Transformer MPC compared to one-step-ahead LSTM networks. Case studies demonstrate a significant fifteen-fold improvement in computational speed compared to one-step-ahead LSTM, although this improvement may vary depending on MPC factors like the lookback window and prediction horizon.
Keyword:
Transformer neural network architecture
Long short-term memory
Model predictive control
期刊
C
IF:
3.9
论文数:
8.1K
被引数:
1.7W
机构
引用论文
分享收藏Physics-constrained deep learning of multi-zone building thermal dynamics基于物理约束的多区域建筑热动力学深度学习.png)
ENERGY AND BUILDINGS
OAAI
ENERGY AND BUILDINGSIF7.1
OAAI分享收藏分享收藏
查看更多内容