基于状态调制注意力机制的区域空气质量时空预测方法及系统
By introducing discrete physical state embedding and state modulation spatial attention mechanisms, the problem of insufficient physical mechanism perception in existing air quality prediction technologies is solved, achieving high-precision prediction of extreme pollution processes and accurate simulation of dynamic spatial transmission, thus improving the robustness and applicability of the model.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- HANGZHOU PONY TECHNOLOGY CO LTD
- Filing Date
- 2026-05-06
- Publication Date
- 2026-07-17
AI Technical Summary
Existing air quality forecasting technologies lack physical mechanism perception and cannot accurately simulate dynamic spatial transport characteristics, resulting in insufficient ability to capture extreme pollution processes. Furthermore, the static map structure cannot adapt to the dynamic directionality and state dependence of atmospheric pollution transport, affecting the accuracy of regional collaborative forecasting.
By introducing discrete physical state embedding and state-modulated spatial attention mechanisms, the physical state perception capability of air pollution is explicitly injected into the deep learning model. The spatial influence weights are dynamically adjusted through the state-modulated spatial attention module to achieve collaborative prediction of pollutant concentrations.
It significantly improves the ability to capture extreme pollution processes, enhances the accuracy of regional collaborative prediction, reduces computational resource consumption, adapts to different regions and pollution scenarios, and strengthens the robustness and real-time performance of the model.
Smart Images
Figure CN122133113B_ABST