In order to improve the service quality of electric power customer service, an intelligent question answering system for electric power customer service is proposed. Extracting and representing important information and context information based on convolutional neural networks (CNN) and bidirectional long short-term memory (BiLSTM), extracting semantic information and further representing the semantic information as a feature vector by combining a BiLSTM network and a collaborative attention mechanism, solving the dependence problem between front and rear words in a long sentence, and obtaining the related feature representation between question pairs; A similarity calculation function is proposed to reconcile the cosine similarity and Euclidean distance to achieve efficient matching of the problem pairs. Taking the power data provided by a power company as an example, the proposed model is verified. The results show that the performance of the proposed model is the best, the precision and recall are 90.96% and 88.63%, respectively, which provides a reference for the development of electric power customer service intelligent service.