Statistical missing data interpolation algorithm for water and soil conservation engineering of overhead transmission lines based on time series characteristics and GRU model

The statistical data of water and soil conservation works for overhead transmission lines often suffer from incompleteness due to variations in project scale. To address this data gap, this study proposes an innovative interpolation algorithm that combines time-series characteristics with a Gated Recurrent Unit (GRU) model to accurately estimate missing values in water and soil conservation statistics. The proposed methodology first analyzes the types of statistical data involved and collects multi-source information to construct the initial dataset. The Isolation Forest algorithm is then employed to detect and remove outliers from the raw data during the preprocessing stage. A novel feature extraction approach integrates self-attention mechanisms into convolutional neural networks to effectively identify and focus on crucial temporal patterns. The developed GRU model, trained on these extracted time-series features, generates reliable predictions for missing data points. Experimental results demonstrate the algorithm's effectiveness, achieving an impressive explained variance score of 0.92 and a Pearson correlation coefficient of 0.94 across different missing data rates. This approach not only ensures data completeness and accuracy but also provides a solid foundation for planning and management decisions in water and soil conservation projects for transmission line infrastructure. The successful implementation of this technique offers a robust solution to a persistent challenge in infrastructure monitoring and maintenance data management.