制冷技术2025，Vol.45Issue(6)：40-46,7.DOI:10.3969/j.issn.2095-4468.2025.06.201
基于张量-时域卷积的制冷量预测方法
Refrigeration Capacity Prediction Method Based on Tensor-Temporal Convolutional Network
张皓鑫 1张欣林 1刘飞天1
作者信息
- 1. 上海碳索能源服务股份有限公司,上海 201108
- 折叠
摘要
Abstract
A refrigeration capacity prediction method based on tensor-temporal convolutional network is proposed to extract latent multi-dimensional features and forecast refrigeration capacity trends.By representing multivariate time-series data in tensor form,this approach captures inter-dimensional correlations through the multi-linear properties of tensors.Temporal convolutional networks are then applied to perform deep learning on the tensor-structured data,effectively modeling both long-and short-term temporal dependencies.Using operational data from a semiconductor cooling system in Xiamen,a multi-dimensional refrigeration capacity prediction dataset was constructed.Compared to conventional time-series forecasting methods,the proposed model achieved a mean absolute percentage error(MAPE)of 1.07%,demonstrating superior accuracy in refrigeration capacity prediction.关键词
制冷系统/张量积/时域卷积网络/负荷预测Key words
Refrigeration system/Tensor product/Temporal convolutional network/Load forecasting分类
通用工业技术引用本文复制引用
张皓鑫,张欣林,刘飞天..基于张量-时域卷积的制冷量预测方法[J].制冷技术,2025,45(6):40-46,7.
