In order to solve the problem of serious consequences caused by untimely processing of 95598 emergency work orders, a parallel hybrid neural network sentiment analysis model for power work order texts based on attention mechanism is proposed to meet the requirements of automatic intelligent reminders for emergency work orders. Firstly, a sentiment analysis dataset for electric power work order texts is generated by manually annotating the emotional urgency of work order texts through sentiment dictionaries and rule sets；Using text pre trained BERT model for text vectorization；Then, use text convolutional neural network（TextCNN）and bidirectional long short term memory（BiLSTM）to extract local and contextual features of the text, respectively, and perform feature fusion；Using attention mechanism to enhance the ability of the model to recognize key information in the fused text features. The calculation results show that the neural net?work model that integrates attention mechanism has better performance in sentiment classification of power work order texts compared toother deep learning models.