The Industrial control systems are widely used in various industries and national critical infrastructure. With the higher security requirements, encrypted protocol communication is gradually used, which produces encrypted industrial control system network traffic, which improves system security while making traditional network traffic detection and analysis technology invalid, thus bringing new network security challenges. Existing work on encrypted traffic identification mainly focuses on the identification of encrypted traffic in information technology networks, and ignores the identification of encrypted traffic in operational technology networks in industrial control systems. Due to the lack of encryption protocol communication data set of industrial control systems with controllable communication behavior, the existing traffic identification models are difficult to effectively extract long-range dependencies of traffic. In this paper, the Tennessee Eastman simulation scenario is constructed based on the OpenPLC and OPC UA protocols for generating the encrypted traffic of industrial control systems, and the industrial control Systems encrypted traffic dataset ICS-ET-2023 is constructed. TransTraffic, a five-layer traffic identification model using packet size features, is constructed based on Transformer encoder and decoder for identifying various operational behaviors in the encryption protocol traffic in the industrial control system network. Evaluation experiments on the ICS-ET-2023 dataset show that the proposed scheme is superior to the state-of-the-art scheme.