Aiming at the problem that the braking distance was too long due to insufficient pedal force of the driver in an emergency, the research took the electric power hydraulic booster braking system as the research object and proposed a method for identifying braking intention based on Hidden Markov Model. The booster motor was controlled to perform normal or emergency braking boost based on the identification of braking intention. The rotation angle, rotation speed and vehicle speed were selected as the identification parameters. The identification parameter data set were divided with the brake intensity as the limit to train the parameters for identifying the normal and emergency braking models, and a recognition model library was built. The driver's braking intention was determined by comparing the log-likelihood estimation of each model library. The results show that the model can accurately identify the braking intention in real-time. When the pedal force was constant, the booster with braking intention recognition control has better braking effect and improves driving safety.