基于多源数据融合与时序预测的智能加药控制方法及系统
The intelligent dosing control method, which integrates multi-source data fusion and time-series prediction, solves the problem of lag in the dosing process of traditional water treatment systems under low temperature and low turbidity conditions. It realizes real-time dynamic control of coagulation reaction and improves the accuracy and safety of the dosing process.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- SHANGHAI PANDA MACHINEGRP CO LTD
- Filing Date
- 2026-04-08
- Publication Date
- 2026-07-17
AI Technical Summary
Traditional water treatment dosing control systems struggle to reflect the microscopic instability of coagulation reactions in real time under low temperature and low turbidity conditions. This results in a lack of observability and physical lag in the dosing process, which can easily lead to substandard effluent and excessive use of chemicals, increasing operating costs and posing risks to water supply safety.
By collecting multi-source data, including raw water intake sensor signals, online flow current meter electrical signals, and real-time underwater image flow, raw water operating conditions, colloidal electrical balance, and floc morphology attribute sets are extracted. A causal mapping sample set is generated and dynamic weighting factor sequence calculation is performed. Combined with time series prediction, a corrective increment for drug replenishment is generated to achieve intelligent decision-making for drug dosing control commands.
It improves upon the problems of missing perception dimensions and appearance deception in traditional control logic, avoids excessive oscillations caused by feedback response deviation, eliminates feedback lag deviation in the dosing process, and improves the accuracy and safety of the dosing process.
Smart Images
Figure CN121978970B_ABST