A novel data cleaning and completion framework for NB-IoT water meters in district metered areas based on purification ratio coefficients

This study presented a systematic methodology for cleaning and completing water consumption data from NB-IoT meters in district metered areas (DMAs). The proposed approach introduced a two-stage anomaly detection method for household meter data, combining threshold-based filtering with differential analysis. A novel intermittent demand analysis framework is developed to identify and process zero-value consumption patterns. For DMA main meter data, a comprehensive preprocessing workflow is established, integrating density-based spatial clustering of applications with noise (DBSCAN) clustering to handle continuous anomalous data segments. The method employs purification ratio coefficients for data completion, considering both temporal patterns and user behavior characteristics. Case study analysis using real DMA data from Langfang City demonstrated the method's superior performance across various missing data scenarios. Compared with traditional approaches, the proposed method achieved significant improvements in accuracy, enhancing prediction precision by 17.72–31.69% across different scenarios. The methodology provides a reliable foundation for smart water management systems while maintaining computational efficiency. This research addresses critical challenges in water supply network data cleaning and offers practical value for intelligent water network management.