Fire control method and system for a plant based on multi-modal data
By combining multimodal data fusion and deep learning with flame simulators and firefighter simulations of real-world scenarios, the accuracy problem of existing fire alarm systems has been solved, enabling accurate early warning and efficient control of fires in large-span factory buildings.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- SHANGHAI FIRE RES INST OF MEM
- Filing Date
- 2026-04-27
- Publication Date
- 2026-07-17
AI Technical Summary
Existing fire alarm systems rely on single-type sensors and lack comprehensive collection and fusion analysis of multi-dimensional environmental parameters. This makes it impossible to provide accurate early warning level judgments in the early stages of a fire, resulting in low accuracy of flame marking and affecting the accuracy of firefighters' scenario construction and fire control.
By monitoring the factory environment through multimodal data, parameters such as temperature, humidity, smoke concentration, and flame radiation intensity are collected. These parameters are then fused with real-time monitoring images to determine the initial warning data combination using deep learning. A flame simulator is used to construct flame markers and warning levels, and simulations are performed based on firefighters' actual combat scenarios and fire-fighting plans to output fire control information.
It improves the accuracy of flame marking and fire control, enables multi-area identification of fire patterns and optimization of fire protection plans, and ensures effective action by firefighters in complex factory fires.
Smart Images
Figure CN122416619A_ABST