The electricity ticket system is an important basis for staff to carry out maintenance plan designation and execution. With the gradual progress of the work of the grid grassroots staff to reduce the burden, the need to simplify the process of electricity ticket invoicing and improve the efficiency of electricity ticket invoicing is increasing. This study first optimizes the scheduling and data capture problem of electricity based on queuing theory. Then, it constructs a recognition model for electric ticket data based on deep learning and combines the CTPN text detection algorithm with the DesNets text recognition algorithm to establish the CDNets algorithm model for recognition optimization. An automated generation model of the blackout ticket was created using the LSTM network. Taking into account the power company’s needs, combined with Intranet intranet software development concepts and Net software technology development and design of the ticket automation system. The AP of this paper’s blackout ticket recognition model under the mixed date of the ticket has a disadvantage of 0.7% compared with YOLOv4, but the reasoning speed is relatively improved by 2 times, and the overall algorithm is more cost-effective. The highest accuracy, F-value, and AP of this paper’s model for stamp data are 95.3%, 97.6%, and 99.2%, respectively. And in the login test and invoicing test, the system in this paper is faster than the traditional system by 298.4ms and 136.7ms, respectively. The electric ticket automation system in this paper is able to improve the automatic generation of electric tickets and the efficiency of electric ticket invoicing, and to enhance the level of lean management at the grassroots level of the grid, with wide adaptability.