Evaluation and imputation of low-quality energy data based on deep learning

The Energy Internet integrates the generation, transmission, storage, conversion, and consumption of various types of energy, achieving efficient utilization and optimized management of energy. During this process, issues like data loss and data transmission errors occur in the measurement, recording, and transmission of data. This paper proposes the use of a data variance statistical algorithm to evaluate data quality and remove low-quality data accordingly. Secondly, it introduces a time series data imputation method based on the Informer model. By considering temporal correlations, this method addresses the evaluation and missing data issues within the Energy Internet. Experimental results demonstrate that the Informer-based time series data imputation method can reduce the error in data imputation and improve the accuracy of imputation data.