Intelligent Error Correction Method for Electric Power Text Based on ELECTRA

In the power industry, text information processing is a crucial task. Whether it's a day-to-day official document release, news writing, or an editorial response to a text related to the power system, a large amount of textual information needs to be processed precisely. With the continuous progress of technology and the expansion of application scope, text correction technology will bring more innovation and development opportunities to the power industry, and is of great significance to facilitate the sustainable progress of the power industry. In this paper, the statistical language model and the neural network-based pre-trained model are fused to train and verify the correction of typos in the power text, and finally the model effect is evaluated through the comprehensive index F1. The results show that the F1 values of the single pre-trained model and the single statistical language model are 0.5871 to 0.4738, respectively, which is lower than the F1 of the fusion model for text typo correction of 0.6354. This fusion method has a good effect on text typo correction in the power industry, and makes a positive impact in ensuring the quality of text related to the power system.