The purpose of this paper is to study and develop an intelligent identification and pre-alarm model of bird damage in electric transmission line(ETL) based on algorithm optimization, so as to solve the security risks caused by bird activities in ETL. In order to achieve this goal, deep learning technology is adopted in the study, and the optimized EfficientNet network is used as the core model. In the aspect of data set construction, a rich data set containing 1600 bird damage images is constructed by combining the resources of on-site shooting images and network images. Through comparative experiments, the performance of the optimized algorithm on ETL bird pest recognition tasks was evaluated compared with traditional Convolutional Neural Networks (CNN) and Long Short Term Memory (LSTM) algorithms. The results show that the precision, recall, and F1 score of this algorithm are significantly better than traditional algorithms. These excellent performance indicators validate the high accuracy and sensitivity of our algorithm in ETL bird pest recognition tasks. This model can achieve high-precision identification of ETL bird pests. The research results are of great significance for improving the safety management level of ETL and reducing power accidents caused by bird damage.