The quality of electricity consumption data significantly influences business model construction, affecting performance and accuracy. This paper proposes a method for analyzing the irreparability of diverse electricity consumption data based on improved data generation technology. Firstly, extracting comprehensive and accurate features from electricity consumption data is difficult due to challenges like low collection frequency and unstable quality. A method for feature extraction is introduced. Secondly, datasets with varying frequencies, time scales, and consumption levels are generated to study the impact of defects on clean data. A method based on the Time GAN model is proposed for reconstructing data while preserving original features and distributions. Finally, a framework for assessing and analyzing the irreparability of electricity consumption data is established. Results show that with an acquisition frequency of 5 min or above, the data can withstand at least 62.9% of defect strength. High-frequency acquisition improves data recoverability by about 10% compared to low-frequency acquisition under the same defect strength.