Accurate prediction of electricity sales plays a positive role in power companies’ reasonable arrangement of power supply plans, scientific optimization of power resource allocation, improvement of power management efficiency, and energy saving and consumption reduction. To enhance the accuracy of electricity sales forecasting for electricity sales companies, this study introduces a new electricity sales forecasting method that utilizes an enhanced Transformer model combined with a sliding window data augmentation technique based on historical electricity sales datasets. The sliding window mechanism generates new training samples by constructing sequences of consecutive data points, thereby enhancing the model’s ability to capture temporal patterns. The simplified Transformer model consists of a single encoder with multi-head self-attention layers and a feedforward neural network. The model’s performance is compared with traditional methods, including a Transformer model without data augmentation and LSTM, using consistent training parameters. The results indicate that the proposed model has high accuracy and reliability in forecasting electricity sales.