Intelligent water quality prediction and regulation method in power plant descaling process
By standardizing and removing anomalies from water quality data from thermal power plants, an enhanced delay vector and covariate weighted matrix are constructed. Combined with local weights and anomaly factors, the problems of single data and poor adaptability to abnormal operating conditions in the descaling process of thermal power plants are solved, and high-precision water quality prediction and dynamic control are achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- HANGZHOU HUADIAN JIANGDONG THERMAL POWER CO LTD
- Filing Date
- 2026-02-10
- Publication Date
- 2026-07-24
AI Technical Summary
Existing technologies for descaling in thermal power plants suffer from problems such as limited data acquisition, insufficient utilization of variable correlation, poor adaptability to abnormal operating conditions, and low prediction accuracy, leading to incomplete descaling or excessive use of chemicals, which increases economic and environmental burdens.
By standardizing and removing anomalies from multi-source water quality monitoring data, an enhanced delay vector is constructed. Combined with local fluctuation suppression, water temperature deviation adjustment, reagent mutation suppression, and time decay weights, a covariate weighted matrix is established. Direction-aware weights and anomaly sensitivity factors are introduced to form a global state vector. The model is trained using Huber loss and heteroscedasticity robust task loss of uncertainty factors.
It improves the accuracy and robustness of water quality prediction for descaling in thermal power plants, provides a reliable basis for dynamic control, ensures the interpretability and causal consistency of prediction results, reduces the impact of outliers and highly uncertain data, and improves the stability and prediction accuracy of the model.
Smart Images

Figure CN121684541B_ABST
Abstract
Citation Information
Patent Citations
CN104318325A
CN110308705A
CN120595735A
WO2024084485A1