一种基于自适应图构建的多设备用能态势时空预测方法
By using an adaptive graph construction method, deep-level connections between devices are automatically extracted, solving the spatial correlation problem in multi-device energy consumption status prediction and achieving high-precision energy consumption status prediction, which is suitable for energy supply and consumption balance scheduling in manufacturing enterprises.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- CHONGQING UNIV
- Filing Date
- 2023-11-24
- Publication Date
- 2026-07-17
AI Technical Summary
Existing equipment energy consumption prediction methods ignore the spatial correlation between equipment in multi-equipment scenarios, resulting in reduced prediction performance. Furthermore, traditional spatiotemporal prediction methods require predefined graph structures that are difficult to adapt to the complex processing flow of equipment in manufacturing workshops.
An adaptive graph construction method is adopted, which optimizes the relationship between devices during the model training process by randomly initializing the graph structure, constructs an adaptive graph neural network, and combines it with a spatiotemporal prediction model to predict the energy consumption status of multiple devices and automatically extract the deep-level connections between devices.
It enables accurate spatiotemporal prediction of energy consumption status of multiple devices without relying on predefined maps, improving prediction accuracy, saving the cost of acquiring predefined maps, and exhibiting strong adaptability and low actual prediction error.
Smart Images
Figure CN117634679B_ABST